Gender and job-related non-formal training: A comparison of 20 countries

Abstract

This article analyses gender differences in the participation in various types of job-related non-formal training in 20 societies and examines the relationship of these gender differences with country-specific institutional settings such as employment protection, family policies and the gender culture. Using data from the Programme for the International Assessment of Adult Competencies (PIAAC) and applying two-step multilevel regression analyses, two main findings are obtained: First, gendered participation clearly differs among training types, with women being less likely to participate in employer-financed training but more likely to participate in non-employer-sponsored training. These gender differences in training participation are crucial because they are likely to shape men’s and women’s career development in different ways, that is, by providing better future career prospects with the current employer for men and with a new employer for women. Second, country-specific settings can reduce gender differences in training participation: in countries with family policies supporting females’ employment (e.g. good coverage of formal childcare and short parental leave), we found a lower training disadvantage of women in employer-financed training. In turn, gender differences in non-employer-sponsored training seem to be lower in countries with less rigid employment protection.

Keywords

Cross-national comparison gender differences gender inequality institutions training

Introduction

Job-related non-formal training (referred to simply as ‘training’ in the following) is the main instrument of employers and employees for adapting skills to changing labour market and job demands.¹ For employers, training makes employees more productive and provides a competitive advantage (Conti, 2005). Employees benefit from training as it is often associated with better job rewards, for example, in terms of wages (Jones et al., 2011).² However, training usually involves monetary investments, which have to be born either by the employer (employer-sponsored training) or by the employee (non-employer-sponsored training (NEST)).³ And these training investments appear to differ among men and women, with a higher participation of women in NEST (Bassanini et al., 2005; Dämmrich et al., 2014) and a lower participation in employer-sponsored training participation (Albert et al., 2010; Georgellis and Lange, 2007; Grönlund, 2011).

However, there is also empirical evidence that women participate to a similar (Albert et al., 2010; Bassanini et al., 2005) or even higher extent (Dämmrich et al., 2014; Kosyakova, 2014) in employer-sponsored training than men. We argue that these inconsistent results regarding gender differences in employer-sponsored training participation might arise due to country-specific institutional setups. Support for this argument can be found in findings from comparative studies. For example, Wozny and Schneider (2014) demonstrated that gender differences in overall training participation are related to countries’ institutional settings, such as family policies and the labour market system. Similarly, Dieckhoff and Steiber (2011) examined institutional packages (via the welfare regime approach), arguing that the modern gender culture and the high, continuous female labour market participation in Nordic countries reduce females’ disadvantage in training participation. Nevertheless, to the best of our knowledge, no comparative study distinguished between employer-sponsored training and NEST activities and explored how gendered participation in these distinct training types varies by specific institutional settings, on one hand, and across welfare regimes, on the other hand.

Drawing on the most recent comparative data for 20 societies from the Programme for the International Assessment of Adult Competencies (PIAAC), this article aims to fill this gap, by addressing the following questions: first, do men and women with comparable characteristics differ in their participation in employer-sponsored training and NEST?⁴ If so, how does this gendered training participation vary among countries? Second, can country-specific characteristics and conventional country classifications contribute to explaining this country variation in gendered training participation? More specifically, we focus on three institutional characteristics: employment protection legislation, family policies (parental leave and childcare) and the gender culture. Given the cross-sectional nature of the PIAAC data, we seek to describe gendered patterns in training participation and their association with country-specific institutions and policies and not to disentangle causal effects.

The knowledge of gender differences in training participation can enhance our understanding of further gender inequalities in the labour market. First, training goes hand-in-hand with higher task complexity of jobs, making also the access to desirable positions – such as supervisory ones – and higher earnings more likely (Tomaskovic-Devey and Skaggs, 2002). However, as we argue, this is likely to hold mainly for employer-sponsored training that strengthens career chances with the same employer. In turn, NEST more likely improves opportunities to find new and/or better jobs (Evertsson, 2004). Hence, gender differences in the participation in both training types might be an important mechanism for the emergence and maintenance of labour-market-related gender inequalities. It may also lead to further ‘cumulative (dis)advantages’ (DiPrete and Eirich, 2006) in the career trajectories of men and women. Second, studying these gender differences among countries and connecting them to their country-specific setups contribute to our understanding of which institutional frameworks foster gender equality.

Gender differences in training participation

In order to develop hypotheses regarding gender differences in employer-sponsored training and NEST participation, we begin with a discussion of the gender bias in skills acquisitions. We then turn to the role of institutional arrangements and rely predominantly on the Varieties of Capitalism approach and welfare regime classifications.

Employers are more likely to invest in training when it develops firm-specific skills. These skills are hardly transferable to other firms and hence bind employees to employers. In turn, NEST is likely to develop more general skills, which are useful for a large number of employers. This type of ‘general’ training thus improves the overall qualification profile of participants and facilitates employees’ mobility between firms (Becker, 1962; Estevez-Abe, 2005).

Participation in both employer-sponsored training and NEST is likely to differ among women and men. The major reason is that women (still) assume the main share of household and childcare duties in most societies (Treas and Drobnic, 2010). These responsibilities make them more likely to interrupt, quit or change their careers compared with men. This, in turn, should influence both employers’ and (female) employees’ training investments.

The human capital theory (Becker, 1985) claims that employers invest in training based on (rational) cost–benefit considerations. As employers have only imperfect information about the (future) productivity of employees, they have to rely on group characteristics such as gender as proxies for individuals’ productivity (Arrow, 1973; Phelps, 1972). Since women (compared with men) tend to have less stable career paths, employers might consider investments into female training as less secure and profitable (Estevez-Abe, 2005).

Hypothesis 1. Women are less likely to participate in employer-sponsored training than men.

Due to the higher volatility of females’ career pathways and the fewer training opportunities provided by employers, women – who want to stay competitive on the labour market – should be more motivated to invest in the acquisition of general skills. These skills are more easily transferable to other employers and hence facilitate job changes and allow for less loyalty to a particular employer. Women’s demand for training might be further strengthened as a consequence of growing women’s career orientation due to, for example, the desire for self-realization on the labour market and financial independence (Gornick and Meyers, 2009). Men, in turn, are likely to participate less in NEST because their greater opportunities to get employer-sponsored training should lead them to focus more on their current careers, on one hand, and limit their interest and/or capability in terms of time and efforts for NEST, on the other hand.

Hypothesis 2. Women are more likely to participate in NEST than men.

Country-specific institutional settings

Comparative studies claim that country-specific institutions tend to shape gender differences in training participation, and these differences are found to vary from country to country (Albert et al., 2010; Dieckhoff and Steiber, 2011; Wozny and Schneider, 2014). Until now, however, only limited frameworks have explained this country variation (Wozny and Schneider, 2014). In the following, we draw on gendered versions of (1) the Varieties of Capitalism approach (Hall and Soskice, 2001) and (2) welfare regime classifications that are based on Esping-Andersen’s (1990, 1999) seminal work. We focus on three country characteristics conceived in order to shape gender differences in training participation: unemployment protection, family policies and the gender culture.

The Varieties of Capitalism approach connects aspects of social protection with skill and production formation and distinguishes two main market economies. The first type is coordinated market economies (CMEs; e.g. Germany, Japan, the Netherlands, Belgium, Sweden, Norway, Denmark and Finland), which rest on complex production strategies that demand high levels of specific skills acquired through training and long-term reciprocal relationships between employers and employees. Certain institutional settings, such as strong employment protection, foster these long-term commitments. In contrast, liberal market economies (LMEs; e.g. United States, United Kingdom and Ireland) are characterized by less complex production strategies, higher market competition and less importance of long-term employer–employee commitments. Coupled with low employment protection, job changes and poaching occur more frequently. Company training serves mainly to train workers in firm-specific skills since the formal educational system is oriented towards general skills. However, due to the higher risk of poaching of trained workers, firm-specific training is less intensive than in CMEs (Dieckhoff et al., 2007; Hall and Soskice, 2001).

How are these institutional characteristics related to gender differences in training participation? We expect that in CMEs (characterized by strong employment protection), employers are more reluctant to invest in the training of women compared with LMEs (characterized by low employment protection). Since employers in CMEs base their product strategies heavily on the firm-specific skills and knowledge of their workers, frequent employment interruptions and/or employer changes are detrimental to their profit. Family-based employment interruptions – which are more frequent for women – are hence more costly for employers in CMEs, due to three main reasons: first, these costs arise as a result of recruiting and training of temporary (replacement) employees and are even higher for training of new employees in case women do not return to their former jobs after childrearing (Estevez-Abe, 2005). Second, even if female employees re-enter their former jobs after career interruptions, returns to training investments diminish due to skill atrophy. Third, re-entry after childbirth is likely to be associated with working part-time, which imposes time restrictions to reap returns to training. This is particularly the case in countries with high employment protection, in which part-time employment is more widespread (Buddelmeyer, Mourre and Ward, 2004). In contrast, in LMEs, employment interruptions – and females’ discontinuous careers, in particular – play a minor role as the labour market is generally more volatile and specific skills are less important. Moreover, the costs of replacing a worker, including firing and hiring, are comparatively lower.⁵

Hypothesis 3. In countries with stronger employment protection (usually CMEs), the disadvantage of women in employer-sponsored training participation is more pronounced.⁶

In turn, we expect the female participation advantage in NEST to be more pronounced in CMEs compared with LMEs. In this context, strong employment protection and a lack of active state support for full employment – which is, for example, the case in the Southern European countries (Italy, Spain and Portugal) – create so-called insider–outsider labour markets. In these labour markets, employees belong to two groups: ‘insiders’ with good and secure jobs and high training provisions and ‘outsiders’ with bad and insecure jobs and fewer training opportunities (Lindbeck and Snower, 1986). Since women more often belong to the latter group (Esping-Andersen, 1999), they might try to improve their chances to access ‘insider’ jobs by investing in training on their own (Blossfeld et al., 2014). Additionally, the anticipated female disadvantage in employer-sponsored training in CMEs (see Hypothesis 3) might compel women to invest more in NEST.

Hypothesis 4. In countries with stronger employment protection (usually CMEs), the advantage of women in NEST participation is more pronounced.

Apart from labour market characteristics, family policies such as childcare and leave arrangements after childbirth are likely to shape gender differences in training participation (Wozny and Schneider, 2014). In countries with more generous formal childcare facilities and shorter parental leave phases, females’ labour force participation is higher (Jaumotte, 2003; see An, 2013 for a literature review).⁷ Higher labour force participation of women is linked, in turn, to a higher female attachment to the labour market. Under these circumstances, employers’ expected risk of losing returns to investments in females’ training is lower; hence, employers should be less deterred to invest in women’s training (Dieckhoff and Steiber, 2011). Empirical support for this thesis comes from Germany, where extensions of parental leave have been found to lower females’ participation in employer-sponsored training (here defined as employers’ initiative to arrange the training; Puhani and Sonderhof, 2011).

Hypothesis 5. In countries offering family policies that support females’ continuous and full-time labour force participation, the disadvantage of women in employer-sponsored training participation is less pronounced.

Gender differences in NEST participation might also be related to family policies. First, the longer the employment interruptions, the higher the depreciation of skills and knowledge, resulting in a stronger need for women to update skills and knowledge (Dieckhoff and Steiber, 2011). Therefore, women should be more likely to participate in NEST in countries with longer parental leave and less generous formal childcare. Second, in order to compensate for the disadvantage in employer-sponsored training (following Hypothesis 5) and to stay competitive in the labour market, (career-oriented) women might invest in training on their own in countries where family policies discourage females’ labour force participation. Weak empirical evidence for this pattern has been found for Germany (Puhani and Sonderhof, 2011).

Hypothesis 6. In countries offering family policies that support females’ continuous and full-time labour force participation, the advantage of women in NEST participation is less pronounced.

A country’s gender culture – that is, country-specific beliefs and norms about women’s and men’s roles in society and about their share of paid and unpaid work – is also likely to impact on gender differences in training participation. In more gender-egalitarian countries, women’s role is not primarily focused on family and childcare responsibilities, but it is extended to paid work in the labour market. Employers’ discrimination against women – for instance, regarding wages – has been found to be lower in these countries (Triventi, 2013), suggesting that employers rest their decisions less on ascriptive characteristics such as gender. In agreement with this observation, we expect that employers in more gender-egalitarian countries are also less likely to discriminate against women with regard to their training investments compared with employers in more traditional ones.

Hypothesis 7. In countries with more gender-egalitarian culture, the disadvantage of women in employer-sponsored training participation is less pronounced.

In a similar vein, in more gender-egalitarian societies, men and women are more equal in terms of labour market participation (Clark et al., 1991) and thus might have more similar incentive structures. This implies a stronger female career attachment to the current job with less investments in ‘external’ training activities.

Hypothesis 8. In countries with a more gender-egalitarian culture, the advantage of women in NEST participation is less pronounced.

Country grouping

Based on the Varieties of Capitalism approach and welfare regime classifications (e.g. Aspalter, 2006; Esping-Andersen, 1990, 1999; Fenger, 2007; Ferrera, 1996), we group the 20 countries included in this article into six regime types: (1) the Liberal, (2) the Nordic, (3) the Central European, (4) the Southern European, (5) the Post-Socialist and (6) the Asian regime. This country grouping, together with indicators used for the respective institutional arrangements (employment protection, family policies and gender cultural orientation), is presented in Table 1.

Table 1.

Country-specific characteristics.

	Firing costs (in weeks of salary)	Strictness of employment protection	Length of parental leave (in weeks)	Parental leave allowance	Percentage of children in formal childcare	Gender culture
Liberal	15.3	0.9	0.0	0.0	34.3	11.1
Ireland	24.0	1.3	0.0	0.0	30.8	10.5
UK	22.0	1.2	0.0	0.0	40.8	17.1
US	0.0	0.3	0.0	0.0	31.4	5.8
Nordic	16.3	2.3	47.9	84.0	48.1	6.2
Denmark	0.0	2.1	32.0	100.0	65.7	5.1
Finland	26.0	2.2	31.6	70.9	28.6	9.5
Norway	13.0	2.3	46.0	100.0	51.3	8.2
Sweden	26.0	2.6	82.0	65.0	46.7	2.0
Central EU	33.5	2.5	32.5	45.9	41.0	16.1
Belgium	16.0	2.0	24.0	41.1	48.4	19.6
Germany	69.0	2.9	48.0	67.0	17.8	15.6
France	32.0	2.4	32.0	33.6	42.0	21.5
Netherlands	17.0	2.8	26.0	42.1	55.9	7.7
Southern EU	33.5	2.6	20.0	15.0	33.4	16.9
Spain	56.0	2.4	0.0	0.0	37.5	12.4
Italy	11.0	2.8	40.0	30.0	29.2	21.4
Post-Socialist	20.0	2.5	96.4	54.6	12.7	26.6
Czech Republic	22.0	3.1	96.0	74.9	2.2	27.4
Estonia	35.0	1.8	62.0	100.0	17.5	17.9
Poland	13.0	2.2	96.0	23.9	7.9	26.6
Russia	17.0	3.1	72.0	40.0	33.0	29.1
Slovakia	13.0	2.2	156.0	34.5	3.0	31.9
Asian	47.5	1.9	52.0	32.6	33.0	32.3
Japan	4.0	1.4	52.0	50.0	28.3	32.2
South Korea	91.0	2.4	52.0	15.2	37.7	32.3

Sources: World Bank (2009), OECD (2011, 2013a, 2013b), An (2013), Lee (2009), Moss (2011, 2012, 2013); own estimations: ESS (2010, 2004); WVS (2010-2014, 2005-2009).

ESS: European Social Survey; WVS: World Values Survey.

Firing costs (in weeks of salary) accounts for costs of advance notice requirements, severance payments, and penalties arising when a redundant worker is terminated. Strictness of employment protection is composed of eight different aspects of strictness of individual dismissals. It ranges from 0 to 6; higher values indicating stricter employment protection. Parental leave allowance is measured as percentage of the countries’ median income for the period of parental leave length. Percentage of children in formal childcare refers to children aged below 3. Gender culture is measured as percentage of agreement with the statement: ‘When jobs are scarce, men should have more right to a job than women.’ Higher values indicate a more traditional gender culture.

LMEs (Ireland, the United Kingdom and the United States) are a quite homogeneous country group with low employment protection, low social benefits and a minimized role of the state. The state does not provide paid parental leave after childbirth. Individuals (and women, in particular) are often forced to stay in employment in order to maintain adequate living standards (Esping-Andersen, 1990, 1999). These characteristics, together with widespread gender-egalitarian values, are likely to lead to low gender differences in training participation.

The countries covered by the CMEs can be divided into five groups. The Nordic countries (Norway, Sweden, Finland and Denmark) are unique in their extensive state policies that focus on universalism and the equalization of employment chances, particularly regarding women’s labour market participation. These countries show a great emphasis on gender egalitarianism with high levels of public childcare that reduce women’s double burden of combining paid and unpaid work. Compared with the other CMEs, employment protection is quite low. Altogether, we expect small gender differences in training participation for this regime type, which is in line with previous research (Dieckhoff and Steiber, 2011).

The next two country groups include the Central European (Belgium, France, Germany and the Netherlands) and Southern European countries (Italy and Spain). Both of these groups are still oriented towards the male breadwinner model with traditional gender values. However, this orientation is fading in both regime types, with the number of childcare facilities increasing and traditional beliefs and norms declining, but more so in Central European countries. Moreover, Southern European countries have rigid labour markets with pronounced insider–outsider structures, disadvantaging women (Blossfeld et al., 2014). For both of these country groups (but particularly for the Southern European one), we expect greater gender differences in training participation compared with the Liberal and Nordic welfare regimes.

Post-Socialist countries (the Czech Republic, Estonia, Poland, Russia and Slovakia) are characterized by their common socialist past in which gender equality was a central ideological goal. Since the fall of the Iron Curtain, however, the combination of work and family obligations has been becoming increasingly difficult. Today, these countries show the lowest coverage of childcare and a pronounced tendency for long career interruptions after childbirth. Furthermore, Table 1 implies that there is a high agreement that males should have the privilege of getting work when jobs are scarce. Along with moderate employment protection, countries of this regime type might retain high gender differences in training participation compared with countries of the other regimes.

Finally, Asian countries (Japan and South Korea) are characterized by a strong family orientation, with a very traditional gender culture and a still underdeveloped family policy that does not support women’s gainful employment. These countries offer a quite lengthy parental leave, and childcare coverage is among the lowest compared with the countries of other regime types. This leads to the expectation that gender differences in training participation are rather high and similar to the Post-Socialist countries.

Data

Our analyses are based on data from the Survey of Adult Skills (Round 1) carried out between August 2011 and June 2012. The survey is part of the PIAAC, coordinated by the Organisation of Economic Co-operation and Development (OECD). The PIAAC data are particularly suitable for our analyses because it provides rich and the most recent information about different types of learning activities (Kirsch and Thorn, 2013). The target population is 16- to 65-year-old individuals in 24 OECD countries (for more details, see OECD, 2013c).

Our analysis includes only 20 countries due to data availability.⁸ The PIAAC sample is restricted as follows: first, the sample is confined to individuals aged below 55 years because training participation strongly decreases in late adulthood (e.g. Becker, 1962). Individuals aged below 20 years are excluded as well, since their training participation can often be considered a part of initial education (e.g. for the dual system in Germany, see Blossfeld and Stockmann, 1999). Second, to ensure that the analyses include only adult learners (and not those who participated in on-the-job training (OJT) as part of the initial education process), we account for the age and the year when the highest educational level was attained. Third, since we aim to consider employers’ investments in training, we limit the analysis to those who are exposed to their employer’s training investments, that is, employed individuals. In order to account for such exposure, we only consider individuals (1) who did participate in training and were employed at any time of training participation and (2) who did not participate in training and are currently employed or have been employed in the last 12 months (the reference period for training participation is the last 12 months prior to the interview date).⁹

After listwise deletion of missing cases in the dependent and independent variables, the final sample includes 39.32 percent of the original PIAAC sample (31,797 men and 34,096 women). Details on the sample selection and the case numbers for each country can be found in Appendix Tables 5 and 6, respectively.

Method

For our empirical investigation, we use a two-step multilevel method. In the first step, we assess gender differences in training participation among countries (Hypotheses 1 and 2) by fitting logistic regression models for each country separately, with the dependent variables being the three training types (level 1). Analyses are weighted with the sample weights provided by the OECD. In the second step, we test whether the country variation in gender differences in training participation is related to institutional settings, by fitting ordinary least square (OLS) linear regressions with the inclusion of the country-specific variables (level 2). The dependent variables are the estimated beta coefficients (from the first step) for females. Therefore, the error term consists of two components: first, the sampling error, which results from the fact that the dependent variable is estimated rather than observed and, second, the residual variance from the step-2 regressions (level-2 error term). To account for heteroskedasticity, the beta coefficients are weighted by their standard errors following the feasible generalized least square approach (FGLS) of Lewis and Linzer (2005).

We use a two-step multilevel method (instead of the widely used one-step multilevel models) for several reasons: First, the two-step approach is better suited for analyses based on a large number of level-1 (individuals) and a limited number of level-2 units (countries; in our case only 20; Franzese, 2005), whereas the estimates of simultaneous one-step multilevel models are sensitive to the number of level-2 units. Especially for logit models, methodological research has revealed a necessity of at least 30–35 level-2 units for accurate estimation of the parameters and standard errors via a one-step multilevel approach (Bryan and Jenkins, 2015). Moreover, one-step multilevel models treat the country effects as random slopes, rendering the validity of results dependent on parametric assumptions. In contrast, two-step approaches calculate country-individual slopes in the first step and compare these afterwards in a second step. Hence, we favour this non-parametric procedure as country-specific slopes may have any distribution. Yet, this comes at the expense of lower statistical efficiency compared to one-step multilevel approaches. Furthermore, because of accounting for both within and between variation, one-step multilevel models calculate smaller standard errors for level-2 variables, which are however underestimated. Two-step approaches, in turn, produce correct and unbiased standard errors (for a formal discussion and exploration of simulation, see Bryan and Jenkins, 2015; also see Austin, 2010; Bowers and Drake, 2005). Finally, while standard one-step multilevel approaches would constrain coefficients of covariates to be equal across countries, two-step approaches are more flexible and robust by allowing them to vary across countries (Heisig, 2011).

Variables

As dependent variables (level 1), we consider only training activities that were reported to be mainly job-related and/or undertaken for job-related reasons. We constructed three binary variables: (1) OJT, (2) employer-financed training (EFT) and (3) NEST. OJT consists of training, instruction or practical experience organized by supervisors or co-workers. EFT refers to open and distance learning, seminars and workshops, courses, or private lessons that are fully or partly paid by employers. Hence, OJT and EFT are considered employer-sponsored training. NEST also refers to open and distance learning, seminars and workshops, courses, or private lessons, albeit without any employers’ monetary investments.

At the individuals level (level 1), the key independent variable female (coded 1 for women and 0 for men) is used to measure gender differences in training participation. We further include several confounding covariates that are likely to affect training participation and vary by gender. Educational level accounts for the higher tendency of more educated individuals to participate in training and for recent positive developments in females’ educational attainment (see Blossfeld et al., 2016). Similarly, we control for abilities measured as competencies in literacy (see, for example, Albert et al., 2010; O’Halloran, 2008). Cohabitation and small children are likely to affect the training participation of men and women differently (e.g. Dieckhoff and Steiber, 2011). Participation in training decreases with increasing age due to shorter time horizons needed for cashing in on training investments (Becker, 1962); age as a proxy for labour market experience further accounts for differences in the career behaviour of men and women.¹⁰ Full-time work is associated with higher training participation and women tend to work full-time less often than men (Dieckhoff and Steiber, 2011). We account for firm size (large firm, and an indicator for missing firm size) because larger firms offer more training opportunities and women often work in smaller establishments (Dieckhoff and Steiber, 2011; O’Halloran, 2008). Finally, we control for working in the public sector. Public sector jobs imply less discriminative practices, which might affect selection into these jobs by gender and gender differences in training participation. Moreover, training opportunities are usually higher in the public sector (e.g. Schömann and Becker, 1995). More information about the construction of the individual-level variables can be found in Appendix Table 7.

At the country level (level 2), we are limited in the number of country-specific indicators that can be tested as independent variables simultaneously due to the restricted number of countries (e.g. Goldthorpe, 1997). Therefore, we modelled indicators referring to the same institutional area (see below for details) into one factor, by performing principal component factor analysis over the respective indicators. Given that the eigenvalues were high enough to indicate one latent concept,¹¹ we calculated the factor scores for each country, resulting in standardized variables with a mean of 0 and a standard deviation of 1 among all countries (Hamilton, 2009).

To test the hypotheses on employment protection (Hypotheses 3 and 4), two indicators are combined into one factor: (1) the strictness of employment protection legislation and (2) the firing costs. Higher values of the factor for employment protection indicate stronger employment protection. For family policies (Hypotheses 5 and 6), three indicators are modelled together as one factor: (1) the length of paid parental leave (in weeks), (2) the level of parental leave allowance for this period (as percentage of the country’s median income) and (3) the percentage of children aged below 3 years in formal childcare.¹² A higher proportion of children in childcare facilitates and encourages females’ continuous and full-time labour force participation, whereas long parental leave length and high parental leave allowance have the opposite effect.¹³ As a result, the two parental leave indicators were reverse-recoded before performing the factor analysis. Higher values of the factor for family policies indicate that family policies support females’ continuous and full-time labour force participation. The hypotheses referring to the gender culture (Hypotheses 7 and 8) are tested by the percentage of individuals having agreed or strongly agreed with the statement, ‘When jobs are scarce, men should have more right to a job than women’.¹⁴ Table 1 provides more information on country-specific variables, as well as their values; correlations of the institutional characteristics are given in Appendix Table 9.

Finally, to examine whether gender differences in training participation vary across welfare regimes, countries were classified into six regime types. The values of the factors for employment protection and family policies as well as the respective regime type can be found in Appendix Table 6.

Results

Descriptive results

A first descriptive overview about overall participation rates and the percentages of women among all participants in the three different types of training is provided in Table 2. The countries are ordered according to their regime type. In general, the highest participation rates can be found in OJT (with 39% of employed individuals having participated during the last 12 months) followed by EFT (34%), while participation is the lowest in NEST (17%).

Table 2.

Participation rates in different types of training and percentage of women among all participants (in %).

	OJT		EFT		NEST
	All	Females	All	Females	All	Females
Liberal	43.47	55.00	33.73	53.00	20.27	57.00
Ireland	39.83	52.41	30.51	49.90	19.54	54.16
UK	45.31	58.39	37.79	56.36	17.20	60.55
US	44.84	52.50	29.95	48.75	27.54	54.68
Nordic	46.18	50.00	50.03	48.00	16.10	52.00
Denmark	46.96	52.17	53.04	51.61	11.51	48.70
Finland	55.70	50.74	51.80	47.59	16.83	55.41
Norway	40.47	45.45	51.64	46.29	13.21	48.62
Sweden	40.16	50.11	41.44	45.90	25.46	52.17
Central European	39.39	49.00	38.96	47.00	13.37	56.00
Belgium	37.88	48.31	37.45	47.34	13.49	55.67
Germany	45.70	47.54	40.91	44.63	16.23	58.18
France	24.85	50.00	30.12	46.20	8.42	55.41
Netherlands	53.70	49.73	50.07	47.74	16.89	54.68
Southern European	28.62	45.00	24.29	44.00	13.29	52.00
Spain	34.58	45.44	29.36	44.83	14.59	52.81
Italy	21.16	45.60	17.93	42.16	11.66	51.79
Post-Socialist	33.77	50.00	26.41	47.00	15.24	58.00
Czech Republic	46.91	45.18	38.42	43.58	14.91	56.48
Estonia	43.43	58.22	34.35	54.06	19.12	62.83
Poland	26.65	42.79	20.60	40.06	13.73	48.62
Russia	17.05	67.04	10.09	65.82	13.09	72.68
Slovakia	25.61	43.88	19.86	43.22	13.59	52.88
Asian	35.90	43.00	26.00	39.00	23.09	52.00
Japan	35.78	43.00	29.08	38.79	19.39	53.72
South Korea	35.99	42.94	23.44	38.78	26.16	50.29
All countries	38.63	49.83	34.34	47.49	16.53	54.85

OJT: on-the-job training; EFT: employer-financed training; NEST: non-employer-sponsored training. Programme for the International Assessment of Adult Competencies (PIAAC, 2012); own calculations.

Participation rates in employer-sponsored training (OJT and EFT) appear to be the highest in the Nordic and the lowest in the Southern European countries. For NEST, the highest participation rates can be found in the Asian countries, while participation is the lowest in the Central European and Southern European countries.

Overall, the results do not indicate a consistent female participation disadvantage in employer-sponsored training. While women tend to participate less than men in employer-sponsored training (both in OJT and in EFT) in Central European, Southern European and Asian countries, they seem to participate more or at a similar level as men in Liberal countries (except in the United States for EFT). The same is true for the two Post-Socialist countries of Estonia and Russia; the remaining countries in this group show in turn a quite strong male participation advantage. Nordic countries have the highest variety in gendered participation, depending on the country and the training type. The findings for NEST indicate that in most countries, women’s participation is higher in this type of training compared with men. Overall, gender differences (to the advantage of females) are lowest in the Nordic, Southern European and Asian countries and highest in Post-Socialist countries.

Multivariate results: individual level

Figure 1 presents the estimated net gender differences in training participation, that is, the mean point estimates of females’ beta coefficients of the country-specific logistic regressions (with 95% confidence interval), given that the individual characteristics are statistically controlled for. A positive beta coefficient indicates a female advantage in training participation, whereas a negative beta coefficient means a female disadvantage. Filled dots denote significant coefficients, empty dots non-significant ones.

Figure 1.

Logistic regression models: probability of females’ participation in OJT, EFT and NEST (conditional coefficients and 95 percent confidence intervals).

The findings in Figure 1 suggest that in half of the countries, no statistically significant gender differences in OJT participation exist. In Ireland, Norway, Spain, Italy, the Czech Republic, Poland, Slovakia and Japan, females are disadvantaged in OJT participation, while in Russia they are advantaged. In turn, in none of the countries, females seem to be (statistically) significantly more likely to participate in EFT than men. Taken together, the results imply gender differences to the disadvantage of women in EFT participation, while participation in OJT is more gender-neutral.

Women are statistically significantly more likely than men to participate in NEST in seven countries. In the remaining countries, the female coefficient – although pointing out a female advantage in participation in almost all countries – does not reach conventional levels of statistical significance.

Bivariate results: country level

Figure 2 displays the bivariate relationship between countries’ gender differences in training participation (x-axis) and the country-specific factors (y-axis) through scatter plots and regression lines from simple OLS regressions. The dependent variables are the beta coefficients for females from the first step of our analysis. Hence, a positive (negative) relationship between the country-level characteristic and the two types of employer-sponsored training (OJT and EFT) indicates lower (higher) participation disadvantages of women for rising values of the macro indicator. A positive (negative) relationship between the country-level characteristic and NEST means higher (lower) participation advantages of women for rising values of the macro indicator.

Figure 2.

Scatterplot of countries: conditional effects of females on the probability of participating in OJT, EFT and NEST, and country-specific institutional settings.

First, the bivariate relationship between gender differences in both OJT and EFT (the two upper rows) and employment protection is quite small, with a correlation of .03 and −.02, respectively. Hence, employment protection seems to not correlate with gender-specific participation in employer-sponsored training. Countries with family policies supporting females’ continuous and full-time labour force participation show a lower participation disadvantage of women in both OJT (corr = .43) and EFT (corr = .52). We further find negative correlations of the gender culture with OJT and EFT, indicating a higher female disadvantage in training participation in more traditional countries. This is even more pronounced for EFT (corr = −.43) compared with that of OJT (corr = −.30).

Second, for NEST, we find moderate correlations for all three indicators. Countries with higher employment protection show even higher training participation for women (corr = .45). In countries with family policies supporting females’ continuous and full-time labour force participation, the female training advantage is less pronounced (corr = −.28), while a more traditional gender culture seems to be associated with higher female participation in NEST (corr = .33).

Multivariate results: country level

Table 3 displays the results of the multivariate analysis that examines the relationship between gender differences in training participation and the country-level variables, net of other country-level covariates. The test statistics for the parameters of the country-level model (level 2) have to be interpreted with caution. First of all, our sample of countries is not a random sample. Second, the size of our sample is small. Third, uncertainty in the dependent variable is expressed by country-specific standard errors. Hence, our interest lies in a multivariate description of parameters and country features among the set of countries under scrutiny. We are not aiming on inferring to all countries conceivable. Thus, we illuminate our hypotheses in the set of countries under study instead of aiming at strict statistical inference.

Table 3.

OLS regressions (FGLS approach): conditional effects of females on the probability of participating in OJT, EFT, and NEST, and country-specific institutional settings (N = 20).

	OJT				EFT				NEST
	M1	M2	M3	M4	M1	M2	M3	M4	M1	M2	M3	M4
Employment protection	.02	.02		.03	.02	.01		.02	.06	.06		.06
Family policies	.08*		.06	.07	.09**		.06	.06	−.04		−.03	−.03
Gender culture		−.06	−.02	−.02		−.08*	−.04	−.04		.04	.04	.03
Constant	−.09**	−.09*	−.09**	−.09*	−.22***	−.22***	−.22***	−.22***	.23***	.23***	.23***	.23***
R ²	.17	.09	.16	.18	.26	.20	.29	.31	.22	.22	.12	.24

OLS: ordinary least square; FGLS: feasible generalized least square approach; OJT: on-the-job training; EFT: employer-financed training; NEST: non-employer-sponsored training.

Programme for the International Assessment of Adult Competencies (PIAAC, 2012); own calculations. The conditional effects of females are obtained from logistic regression analyses for every country separately (see also Figure 1 and Appendix Tables 10 to 12).

Significance level: *p < .1; **p < .05; and ***p < .01.

For OJT, the findings from the bivariate examination are supported: countries with higher employment protection (M1, M2 and M4), countries with family policies oriented towards females’ labour force participation (M1, M3 and M4) and countries with less traditional gender culture (M2, M3 and M4) show lower participation disadvantages for women in OJT. The coefficients for employment protection and the gender culture are, however, not significant (neither statistically nor substantially), and the variance explanation of these two indicators is fairly low (see M2: R² = .09). Moreover, when controlling for family policies, the coefficient for the gender culture reduces even more (M3 and M4). Altogether, it seems that family policies matter most for gender differences in OJT participation (note statistically significant coefficient in M1).

Small, insignificant coefficients for employment protection also indicate the low importance of this macro characteristic for gender differences in EFT participation (M1, M2 and M4). However, when controlling for employment protection, the coefficients for both gender culture and family policies become statistically significant (M1 and M2). It seems that the disadvantage of females is lower in countries with more advanced family policies and less traditional gender culture. In models 3 and 4, in which both of these variables are included, neither of the coefficients is statistically significant. However, this might be due to the high correlation between both variables, which generally dissuades researches from simultaneous controlling for both of them. The overall high gains in variance explanation in the full model (M4) on EFT participation compared to OJT participation implies that gender differences in EFT might be particularly contingent on institutional setups.

For NEST, the bivariate results are supported: in countries with higher employment protection (M1, M2 and M4), with family policies not supporting women’s continuous and full-time employment (M1, M3 and M4) and with more traditional gender culture (M2, M3 and M4), women participate in NEST even more than men. The strength of the coefficients and the variance explanation further suggest that employment protection matters more for explaining gender differences among countries compared with family policies and the gender culture (although none of the coefficients reaches statistical significance).¹⁵

Finally, Table 4 indicates that – compared with the Nordic countries – Southern European and Asian countries exhibit a higher participation disadvantage of females in OJT, while Liberal and Central European countries show lower levels. In turn, the participation disadvantage of females in EFT is notably higher in Southern European, Post-Socialist, and Asian countries than in Nordic countries. Overall, it seems that Nordic and Central European countries are characterized by the lowest gender differences compared with the other regime types. For NEST, it is the Post-Socialist countries (followed by Central European countries) that display the highest gender differences in training participation compared with the Nordic countries. Latter ones, together with Liberal countries, are characterized by the lowest gender differences. Nevertheless, high standard errors do not allow any firm conclusion but more a description of trends.

Table 4.

OLS regressions (FGLS approach): conditional effects of females on the probability of participating in OJT, EFT and NEST among country groups.

	OJT	EFT	NEST
Nordic (ref.)
Liberal	.01	−.07	−.00
Central European	.08	.03	.17
Southern European	−.15	−.18	.04
Post-Socialist	−.07	−.15	.24**
Asian	−.13	−.22	.07
Constant	−.06	−.14*	.12
R ²	.17	.32	.35

OLS: ordinary least square; FGLS: feasible generalized least square approach; OJT: on-the-job training; EFT: employer-financed training; NEST: non-employer-sponsored training.

Significance level: *p < .1; **p < .05; and ***p < .01.

Discussion

Using most recent comparative data for 20 societies from the PIAAC study, we aimed to explore (1) gender differences in participation in employer-sponsored training and NEST and (2) the variation in these gendered training participation patterns among regime types and the relation of country-specific characteristics – employment protection legislation, family policies and the gender culture – with these gendered training participation patterns.

We assumed employers to be more likely to invest in male employees’ training because employer-sponsored training is designed to train firm-specific skills that bind employees to the company, whereas women, due to their more frequent career interruptions, imply a less secure investment in this regard (Hypothesis 1). Rather women seem to have higher incentives and/or a greater demand to invest in NEST, due to their more volatile career pathways that make general skills more important to find new/better jobs (Hypothesis 2). Our findings reveal that gender-specific participation varies notably depending on the training type, with women being less likely to participate in EFT overall (patterns favouring Hypothesis 1) but more likely to participate in NEST (patterns favouring Hypothesis 2). For OJT, however, we do not find any female disadvantage (patterns disfavouring Hypothesis 1). While EFT is usually organized as distance or external training, OJT is usually organized as in-house or internal training. Women hence only seem to experience a training disadvantage in the more costly, external training type.

Moreover, we found notable country variation in the gendered training participation. Countries with family policies supporting females’ continuous and full-time labour force participation and countries with more gender-egalitarian culture demonstrate a lower female participation disadvantage in OJT (albeit not statistically significant) and EFT (patterns favouring Hypotheses 5 and 7). This pattern is further supported when looking at different regime types: Nordic countries (which seek to increase women’s labour force participation and are characterized by a pronounced gender-egalitarian culture) imply the lowest female disadvantage in EFT participation. In turn, this disadvantage tends to be notably higher in Southern European, Post-Socialist and Asian countries. These three regime groups are characterized by more traditional gender cultures and less developed family policies. The gender differences in OJT participation by regime type are similar, with greatest disadvantage of women in Southern European and Asian countries.

We did not find the expected higher female disadvantage in employer-sponsored training participation in countries with higher employment protection (patterns disfavouring Hypothesis 3). Rather, there seems to be nearly no association of employment protection with gendered training participation in OJT and EFT. Due to the high importance of specific skills in countries with high employment protection (usually CMEs), employers might tend to invest in the training of all of their employees independent of their gender. Another possible reason might lie in the decreasing length of females’ interruptions due to childcare during the last decades (Del Boca and Pasqua, 2005). Hence, the returns to females’ training in case of shorter employment interruptions might merely be slightly postponed, but not completely lost.

Regarding gender differences in NEST participation, our findings (although not statistically significant) indicate that women invest even more in training participation on their own in countries with strong employment protection, probably because they seek to improve their chances of accessing ‘insider’ jobs (patterns favouring Hypothesis 4). Countries with less traditional gender culture show a more equal training participation of men and women (patterns favouring Hypothesis 8). In line with that, we found that Nordic and Liberal countries incorporate the lowest gender differences in NEST – and these countries are characterized by low employment protection and a pronounced gender-egalitarian culture. In turn, Post-Socialist countries show the highest gender differences, meaning that women participate in NEST even more often – probably to compensate for their pronounced training participation disadvantage in EFT. Furthermore, we found a lower female advantage in countries with more advanced family policies, patterns favouring Hypothesis 6.

From this summary of findings, we may conclude that women face a double disadvantage in most countries. First, they are disadvantaged in terms of their participation in more costly EFT, which is oriented towards specific skills formation and might be particularly valued by the (current) employer. Second, women – maybe partly due to their disadvantage in employer-sponsored training participation – invest more in their own training (i.e. to acquire more general skills) compared with men. Consequently, men are likely to have better career prospects with their current employer because of higher accumulated specific skills. Nevertheless, women’s higher participation in NEST and thus their higher investments in general skills might improve their opportunities of finding new and/or better jobs with other employers (Evertsson, 2004). However, this also means that women’s careers are more precarious, with lower individual labour market security. Finally, the finding that family policies supporting women’s employment and less traditional gender cultures are linked to a lower female disadvantage in employer-sponsored training indicates that strengthening women’s labour market participation and reducing their employment interruptions after childbirth may lessen gender inequalities in the labour market in a broader sense.¹⁶

This study has several limitations. First, our results are limited to the 20 countries under study and cannot be easily generalized to further countries. Beyond low sample size, the non-random choice of countries and uncertainty in dependent variables do not allow any strict statistical generalization but rather can only serve as a multivariate description of the countries under scrutiny. However, by including countries from different welfare state regimes (e.g. Esping-Andersen, 1999) and from the west to the east, our analyses cover a broad heterogeneity of societies. Second, because we are restricted to the use of cross-sectional data, we cannot identify causal effects between country-specific institutional arrangements and gender differences in training participation (which would require longitudinal data at the micro and macro level). We can only identify correlation patterns between institutional constellations and participation in different training types. Third, while participation in training has been found to improve other labour market outcomes, such as wages (e.g. Jones et al., 2011; Tomaskovic-Devey and Skaggs, 2002) and career opportunities (Blossfeld et al., 2014), it would be promising to study the differences in employers’ rewards between employer-sponsored training and NEST. In this regard, we highlight the importance of employers’ investments associated with the specific skills formation, which would apparently increase wage returns. In turn, general skills are less likely to lead to such benefits, thereby resulting in a further accumulation of labour market disadvantages for females.

Footnotes

Appendix 1

Appendix 2 Acknowledgements

Earlier versions of the article were presented at the ISA RC28 conference ‘Old and new social divides: social stratification research in the 21th century’ in Budapest (Hungary) in May 2014 and at the conference ‘How do educational systems shape educational inequalities’ in Luxembourg (Luxembourg) in July 2014. We thank the participants at these events for helpful discussions. We are also grateful to Claudia Buchmann and Jan Skopek and five anonymous reviewers for insightful and constructive comments.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by an Advanced Grant of the European Research Council (ERC) to Hans-Peter Blossfeld (Call details ERC-2010-AdG, SH2, Project-ID 269568).

Notes

References

Akgunduz

Plantenga

(2012) Labour market effects of parental leave in Europe. Cambridge Journal of Economics 37(4): 845–862.

Albert

García-Serrano

Hernanz

(2010) On-the-job training in Europe: Determinants and wage returns. International Labour Review 149(3): 315–341.

M-Y

(2013) Institutional options, cultural orientation and mothers’ work and family reconciliation in Korea. Asian Social Work and Policy Review 7: 157–174.

Arrow

(1973) Higher education as a filter. Journal of Public Economics 2: 193–216.

Aspalter

(2006) The East Asian welfare model. International Journal of Social Welfare 15(4): 290–301.

Austin

(2010) Estimating multilevel logistic regression models when the number of clusters is low: A comparison of different statistical software procedures. The International Journal of Biostatistics 6(1): Article 16.

Bassanini

Booth

Brunello

. (2005) Workplace training in Europe. IZA Discussion Paper no. 1640, June 2005. Bonn: IZA.

Becker

(1962) Investment in human capital: A theoretical analysis. Journal of Political Economy 70(5): 9–49.

Becker

(1985) Human capital, effort, and the sexual division of labor. Journal of Labor Economics 3(1): 33–58.

10.

Blossfeld

Stockmann

(1999) The German dual system in comparative perspective. International Journal of Sociology 28(4): 3–28.

11.

Blossfeld

H-P

(2016) Changes in educational inequality in cross-national perspective. In: Shanahan

Mortimer

Johnson

(eds) Handbook of the Life Course, vol. II, Heidelberg: Springer, pp. 223–247.

12.

Blossfeld

H-P

Kilpi-Jakonen

Vono De Vilhena

. (2014) Adult Learning in Modern Societies: An International Comparison from a Life-Course Perspective. Cheltenham; Northampton, MA: Edward Elgar.

13.

Boeren

(2011) Gender differences in formal, non-formal and informal adult learning. Studies in Continuing Education 33(3): 333–346.

14.

Bowers

Drake

(2005) EDA for HLM: Visualization when probabilistic inference fails. Political Analysis 13(4): 301–326.

15.

Bryan

Jenkins

(2015) Multilevel modelling of country effects: A cautionary tale. European Sociological Review. Epub ahead of print 8 May. DOI: 10.1093/esr/jcv059.

16.

Buddelmeyer

Mourre

Ward

(2004) The determinants of part-time work in EU countries: Empirical investigations with macro-panel data. IZA DP No. 1361, October 2004. Bonn: IZA.

17.

Clark

Ramsbey

Adler

(1991) Culture, gender, and labor force participation: A cross-national study. Gender and Society 5(1): 47–66.

18.

Conti

(2005) Training, productivity and wages in Italy. Labour Economics 12(4): 557–576.

19.

Dämmrich

Vono De Vilhena

Reichart

(2012) Participation in adult learning in Europe: The impact of country-level and individual characteristics. In: Blossfeld

H-P

Kilpi-Jakonen

Vono De Vilhena

. Adult Learning in Modern Societies: An International Comparison from a Life-Course Perspective. Cheltenham; Northampton, MA: Edward Elgar, pp. 29–55.

20.

Del Boca

Pasqua

(2005) Social Policies and Employment of Married Women in Europe. Turin: University of Turin.

21.

Dieckhoff

Steiber

(2011) A re-assessment of common theoretical approaches to explain gender differences in continuing training participation. British Journal of Industrial Relations 49(S1): 135–157.

22.

Dieckhoff

Jungblut

J-M

O’Connell

(2007) Job-related training in Europe: Do institutions matter? In: Gallie

(ed.) Employment Regimes and the Quality of Work. Oxford: University Press, pp. 77–103.

23.

DiPrete

Eirich

(2006) Cumulative advantage as a mechanism for inequality: A review of theoretical and empirical developments. Annual Review of Sociology 32(1): 271–297.

24.

Esping-Andersen

(1990) The Three Worlds of Welfare Capitalism. Cambridge: Polity Press.

25.

Esping-Andersen

(1999) Social Foundations of Postindustrial Economies. Oxford: Oxford University Press.

26.

Estevez-Abe

(2005) Gender bias in skills and social policies: The varieties of capitalism perspective on sex segregation. Social Politics: International Studies in Gender, State & Society 12(2): 180–215.

27.

Evertsson

(2004) Formal on-the-job training: A gender-typed experience and wage-related advantage? European Sociological Review 20(1): 79–94.

28.

Fenger

(2007) Welfare regimes in Central and Eastern Europe: Incorporating post-communist countries in a welfare regime typology. Contemporary Issues and Ideas in Social Sciences 3(2): 1–30.

29.

Ferrera

(1996) The ‘Southern Model’ of welfare in social Europe. Journal of European Social Policy 6(1): 17–37.

30.

Franzese

(2005) Empirical strategies for various manifestations of multilevel data. Political Analysis 13(4): 430–446.

31.

Georgellis

Lange

(2007) Participation in continuous, on-the-job training and the impact on job satisfaction: Longitudinal evidence from the German labour market. The International Journal of Human Resource Management 18(6): 969–985.

32.

Goldthorpe

(1997) Current issues in comparative macrosociology: A debate on methodological issues. Comparative Social Research 16: 1–26.

33.

Gornick

Meyers

(2009) Gender Equality: Transforming Family Divisions of Labor. New York: Erik Olin Wright.

34.

Grönlund

(2011) On-the-job training - A mechanism for segregation? Examining the relationship between gender, occupation, and on-the-job training investments. European Sociological Review 28(3): 408–420.

35.

Hall

Soskice

(2001) Varieties of Capitalism: The Institutional Foundations of Comparative Advantage. Oxford: University Press.

36.

Hamilton

(2009) Principal Components, Factor, and Cluster Analysis. Belmont: Brooks; Cole.

37.

Heisig

(2011) Who does more housework: Rich or poor? A comparison of 33 countries. American Sociological Review 76(1): 74–99.

38.

Jaumotte

(2003) Female labour force participation: Past trends and main determinants in OECD countries. OECD Economics Department Working Papers no. 376, 12 December 2003. Paris: OECD Publishing.

39.

Javornik

(2014) Measuring state de-familialism: Contesting post-socialist exceptionalism. Journal of European Social Policy 24(3): 240–257.

40.

Jones

Kalmi

Kauhanen

(2011) The effects of general and firm-specific training on wages and performance: Evidence from banking. Oxford Economic Papers 64(1): 151–175.

41.

Kirsch

Thorn

(2013) Preface: The programme for international assessment of adult competencies: An overview. In: OECD (eds) Technical Report of the Survey of Adult Skills (PIAAC). Paris: OECD Publishing, pp. 1–22.

42.

Kosyakova

(2014) Job-related adult learning in the Russian Federation: More educational opportunities without an equalization effect. In: Blossfeld

H-P

Kilpi-Jakonen

Vono De Vilhena

. (eds) Adult Learning in Modern Societies: An International Comparison from a Life-Course Perspective. Cheltenham; Northampton, MA: Edward Elgar, pp. 140–161.

43.

Lane

Myant

(2007) Varieties of Capitalism in Post- Communist Countries. London: Palgrave.

44.

Lee

(2009) Gender equality in reconciling work and childcare in South Korea. In: Cornell Law school Inter-University graduate student conference papers 17, Available at: http://scholarship.law.cornell.edu/lps_clacp/17 (accessed 2 December 2014).

45.

Lewis

Linzer

(2005) Estimating regression models in which the dependent variable is based on estimates. Political Analysis 13(4): 345–364.

46.

Lindbeck

Snower

(1986) Wage setting, unemployment, and insider-outsider relations. American Economic Review 76(2): 235–239.

47.

Moss

(2011) International Review of Leave Policies and Related Research 2011. London. Available at: http://www.leavenetwork.org/archive_2005_2009/annual_reviews (accessed 2 December 2014).

48.

Moss

(2012) International Review of Leave Policies and Related Research 2012. London. Available at: http://www.leavenetwork.org/archive_2005_2009/annual_reviews (accessed 2 December 2014).

49.

Moss

(2013) International Review of Leave Policies and Related Research 2013. London. Available at: http://www.leavenetwork.org/archive_2005_2009/annual_reviews (accessed 2 December 2014).

50.

OECD (2011) OECD Reviews of Labour Market and Social Policies: Russian Federation. Paris: OECD Publishing.

51.

OECD (2013a) Family Database. Paris: OECD Publishing, Available at: www.oecd.org/els/social/family/database (accessed 2 December 2014).

52.

OECD (2013b) OECD Indicators of Employment Protection. Annual time series data 1985–2013. Paris: OECD Publishing.

53.

OECD (2013c) Technical Report of the Survey of Adult Skills (PIAAC). Paris: OECD Publishing.

54.

O’Halloran

(2008) Gender differences in formal on-the-job training: Incidence, duration, and intensity. Labour 22(4): 629–659.

55.

Phelps

(1972) The statistical theory of racism and sexism. The American Economic Review 62(4): 659–661.

56.

Puhani

Sonderhof

(2011) The effects of parental leave extension on training for young women. Journal of Population Economics 24(2): 731–760.

57.

Ray

Gornick

Schmitt

(2010) Who cares? Assessing generosity and gender equality in parental leave policy designs in 21 countries. Journal of European Social Policy 20(3): 196–216.

58.

Saxonberg

(2013) From defamilialization to degenderization: Toward a new welfare typology. Social Policy & Administration 47(1): 26–49.

59.

Schömann

Becker

(1995) Participation in further education over the life course: A longitudinal study of three birth cohorts in the Federal Republic of Germany. European Sociological Review 11(2): 187–208.

60.

Tomaskovic-Devey

Skaggs

(2002) Sex segregation, labor process organization, and gender earnings inequality. The American Journal of Sociology 108(1): 102–128.

61.

Treas

Drobnic

(2010) Dividing the Domestic: Men, Women & Household Work in Cross-National Perspective. Stanford, CA: Stanford University Press.

62.

Triventi

(2013) The gender wage gap and its institutional context: A comparative analysis of European graduates. Work, Employment & Society 27(4): 1–18.

63.

World Bank (2009) Doing Business. Washington, DC: The World Bank.

64.

Wozny

Schneider

(2014) A matter of degree: The continuing training gap for women in Europe. Socio-Economic Review 12(2): 353–379.