Women and young workers are two of the most prominent groups

Transcription

1 International Labour Review, Vol. 157 (2018), No. 1 Occupational segregation by hours of work in Europe Theo SPARREBOOM* Abstract: This article quantifies the levels of occupational segregation between part-time and full-time work using data from the European Labour Force Survey for 15 European countries. It also attempts to identify some of the determinants of segregation through regression analysis using three groups of indicators (quantity of employment, quality of employment and institutional factors). Occupational segregation by hours of work is generally higher for males than for females and is also higher for young workers versus adult workers. It is also found that segregation for men is correlated with the quantity of employment, while variables from all three groups of indicators are important for adult women and young workers. Women and young workers are two of the most prominent groups facing systematic disadvantages in the labour market. Female labour force participation rates, for example, are lower than male rates in virtually all countries, and women often receive lower wages. In a sample of countries used in the Global Wage Report 2014/15, the gender wage gap ranged from 35.8 per cent in the United States to 4.0 per cent in Sweden (ILO, 2015a), while in the European Union (EU) women s gross hourly earnings were on average 16.4 per cent below those of men in The position of youth in labour markets around the world is reflected in youth unemployment rates which are usually at least twice as high as adult unemployment rates but also in job quality indicators (ILO, 2013). Occupational segregation by sex accounts for a significant part of the observed wage gap between men and women (World Bank, 2011), 2 and * International Labour Office; sparreboom@ilo.org. The author would like to acknowledge the valuable research assistance provided by Pinar Hosafci, ILO consultant. Responsibility for opinions expressed in signed articles rests solely with their authors, and publication does not constitute an endorsement by the ILO. 1 See [accessed 4 January 2018]. 2 Other factors that have been put forward by researchers include the undervaluation of women s work, workplace characteristics and the overall wage structure in a country, which may reflect wage-setting mechanisms designed with a focus on workers in male-dominated sectors (see ILO, 2015a, and the references therein for overviews). Copyright International Labour Organization 2018 Journal compilation International Labour Organization 2018

2 66 International Labour Review it is also an important factor in explaining part-time pay penalties. Manning and Petrongolo (2008) showed that, in the United Kingdom, the part-time pay penalty became very small if occupational differences between part-timers and full-timers were controlled for. Based on a double decomposition of the gender wage gap (between men and women employed full time, as well as between full-time and part-time working women), Matteazzi, Pailhé and Solaz (2013) found that labour market segregation was more important than the prevalence of part-time employment per se in explaining gender wage gaps. This article quantifies levels of occupational segregation between full-time and part-time work and attempts to identify some of the determinants of this segregation. In the first part of the analysis, segregation between full-time and part-time work is measured for all workers, as well as within subgroups of the employed population (men, women, youth and adults) for 15 European countries. The extent to which segregation is driven by constituent subgroups is also analysed. We show that occupational segregation by hours of work is generally higher for males than for females, thereby suggesting that part-time pay penalties may be more considerable for men than for women, and we show that occupational segregation is higher for young workers than for adult workers. We also use regression analysis across countries to identify factors that drive segregation by hours of work. Three groups of variables are considered, which are related to (1) quantity of employment, (2) quality of employment, and (3) institutional factors. Segregation by hours of work for subgroups of the employed population is found to be driven by different factors. In particular, segregation for men is correlated with the quantity of employment, while variables from all three of the abovementioned groups are important for adult women and young workers. The remainder of this article is organized as follows. Literature related to segregation by hours of work is reviewed in the first section, which also provides a discussion of some of the theoretical background related to segregation between part-time and full-time work. The second section documents such segregation by gender and age groups across European countries, and the third section shows an empirical analysis of the drivers of segregation by hours of work. Finally, the conclusions are presented in section four. Literature review: Explaining segregation by hours of work Part-time work is an important non-standard form of employment, given that the standard employment relationship is often understood as full-time work (ILO, 2015b). Similar to other forms of non-standard employment, parttime work has become more common over the past several decades. In the European Union of 15 Countries (EU15), the share of part-time employment in total employment increased from 17.5 per cent in 2000 to 21.4 per cent

3 Occupational segregation by hours of work in Europe 67 in 20 and reached 22.9 per cent in For women, the share rose from 33.3 per cent in 2000 to 38.0 per cent in 2014, while the respective figures for men were 5.7 and 9.8 per cent. There are few systematic studies on the determinants of segregation with respect to hours of work. Much more research has been concentrated on the incidence of part-time work (Buddelmeyer, Mourre and Ward, 2008; O Reilly and Fagan, 1998) and on wage penalties and other aspects of the quality of part-time work (Bardasi and Gornick, 2008; Manning and Petrongolo, 2008; Matteazzi, Pailhé and Solaz, 2014; OECD, 20). For example, Buddelmeyer, Mourre and Ward (2008) found that cyclical variables exert a negative effect on part-time employment in the short run, while institutions and other structural factors such as changes in legislation are important in explaining the volume of part-time employment over the long term. The 20 Employment Outlook by the Organisation for Economic Co-operation and Development (OECD) demonstrated that the part-time penalty is reflected in a wider range of indicators, including earnings potential, union membership and job security. At the same time, however, it showed that part-time jobs also carried a premium in terms of control of working time and health and safety risks (OECD, 20). An evaluation of the impact of legislation on equal treatment for part-time workers in the same study suggested that legislation was associated with an increased likelihood of having a permanent contract for men and women and an increased likelihood of participation in training for men, thus reducing the job quality gap between full-time and part-time workers, but only in countries with tight labour market conditions (ibid., box 4.3, pp ). Occupational distributions are often related to age, and occupational segregation among young people tends to be greater than the segregation among their adult counterparts. The British Equality and Human Rights Commission showed that in 20 both young men and young women in the United Kingdom were over-represented in elementary occupations and sales jobs and were under-represented in managerial, professional and associate professional positions (EHRC, 2011). 4 Many young people tended to combine education with low-paid, low-quality jobs on a part-time, temporary basis. However, compared with men, women continued to be under-represented in better-paying, higher-status managerial and professional occupations in later years. Andrews, Bradley and Stott (2004) used British careers service data on young people s occupational preferences to determine segregation in the youth training market ( pre-labour market segregation ), and used information on occupational destinations after young people successfully entered the labour market to calculate segregation in the competitive job market ( post-labour market segregation ). Those authors found that in 3 Eurostat, [accessed 4 January 2018]. 4 Similar evidence is available for Australia. A recent survey found that nearly two-fifths of students aged worked part time and that approximately 87 per cent of part-time workers aged worked in low-skilled occupations such as elementary clerical, sales and service jobs (Abhayaratna et al., 2008).

4 68 International Labour Review the United Kingdom, occupational segregation in the pre-labour market was significantly higher than occupational segregation in the post-labour market. In other words, there was a mismatch between workers preferences before they entered the labour market and the jobs they actually obtained. These differences were attributed to (1) the different treatment of young people by their families, schools and peers, according to which young people were being pre-sorted into occupations that fit traditional gender stereotypes; and/ or (2) young people s own perceptions of gender roles in the labour market. Antecol and Cobb-Clark (20) argued that entry into male-dominated fields of study and male-dominated occupations was related to an individual s perception of having male traits, such as the willingness to work hard, impulsiveness and the tendency to avoid problems. These authors contended that such non-cognitive traits provided an important, though incomplete, explanation for segregation in both young people s educational fields and their occupational choices. Tijdens (2002) distinguished four regimes, or models, which explain to varying degrees whether a woman works part time or full time. The first model reflects the supply-side nature of part-time work (the gender roles regime ), according to which the probability that an individual is engaged in part-time employment depends on gender roles in the family. The remaining three models the secondary labour market regime, the optimal staffing regime and the responsive firms regime reflect the demand side. According to the secondary labour market regime, firms tend to create low-wage, part-time jobs with high turnover, while according to the optimal staffing regime, firms create part-time jobs in response to fluctuations in the workload, for example because of time-related demand for services. Finally, according to the responsive firms regime, firms are responsive to workers demands for reduced hours in their jobs. These models also suggest that part-time work is likely to be more widespread in certain occupations (and industries) than in others and that the nature of part-time work is such that segregation between full-time and parttime work can be expected (Sparreboom, 2014). Both the gender roles regime and the responsive firms regime suggest that there is segregation by hours of work along lines that are similar to segregation by sex. If, according to neoclassical theory, the family is viewed as a production unit that can benefit from specialization, then such specialization results in differential choices for working time, that is, a preference for part-time work, which helps to explain segregation by sex. However, occupational choices could also be affected by specialization through linkages between occupations and the need to travel, the availability to work overtime, and so on. According to the responsive firms regime, segregation by sex or by hours will be reinforced to the extent that firms are likely to be more responsive to workers needs with regard to part-time work in workplaces where women (or part-time workers) are already well represented. The optimal staffing regime is also more relevant for certain industries and occupations than for others, as it was stated by the OECD (20), while the

5 Occupational segregation by hours of work in Europe 69 nature of the secondary labour market regime (emphasizing low-wage employment) suggests that there is segregation away from occupations that involve greater skills (and better pay). Segregation according to the abovementioned regimes is also related to the characteristics of both the actual part-time jobs and the workers, such as the number of hours worked and the preference to work more hours. In Norway, for example, involuntary part-time employment is strongly associated with secondary labour market characteristics, while voluntary, short parttime work allows employers more flexibility in low-skill service industries (Kjeldstad and Nymoen, 2012). Segregation by hours of work in Europe Data and methodology The main data source used in this article is the Eurostat European Labour Force Survey (ELFS) for the year 2011, and we consider 15 countries. 5 The key advantages of the ELFS are its consistent measurement across all countries covered by the survey and the sample size, thereby allowing for a precise measurement of the characteristics of small subgroups in the employed population. 6 The Duncan and Duncan index of dissimilarity (ID) is used as the measure of segregation. The ID is defined as half the sum of the absolute differences, over all groups of occupations, between proportions of two groups of workers (part-time/full-time): ID = (1/2) Σ ABS (PT i /PT FT i /FT) where ABS is the absolute value; PT i /PT is the proportion of part-time workers in occupation i; FT i /FT is the proportion of full-time workers in occupation i, and Σ is the summation over all occupations i. We take the measured number of usual weekly hours of work 7 and, following van Bastelaer, Lemaître and Marianna (1997) and Sparreboom (2014), we use a cut-off of 30 hours to define part-time work. The index can generally be interpreted as the minimum proportion of either group that has to change categories in order to make the relative frequency distributions equal (Mulekar, Knutson and Champanerkar, 2008, p. 2099). One advantage of the ID is that it is not directly dependent on the relative volume of employment of any group, therefore allowing for crosscountry comparisons (Charles and Grusky, 1995). 5 These countries are Austria, Belgium, Denmark, Finland, France, Greece, Ireland, Italy, Luxembourg, the Netherlands, Norway, Portugal, Spain, Sweden and the United Kingdom. With the exception of Norway, they are all Member States of the EU before its enlargement in The average number of observations in the employed population is more than 0,000 per country. 7 The number of usual hours of work (different for each worker) is taken from the data and compared with the cut-off (30 hours per week).

6 70 International Labour Review Descriptive statistics In our sample of countries, the average incidence of part-time work is 18.7 per cent, although the average is much higher for women (28.9 per cent) than it is for men (.2 per cent) and is much higher for youth (33.6 per cent) than it is for adults (16.8 per cent). The large differences in incidence reflect the different roles of part-time work for subgroups of the employed population and the extent to which part-time work has been accepted. In several countries in northern Europe (Belgium, Ireland, the Netherlands, Norway and the United Kingdom), the incidence of part-time work exceeds 20 per cent. In the Netherlands, the majority of employed women work part time, and in four countries (Denmark, Ireland, the Netherlands and Norway) the majority of employed youth work only part time. In Greece and Portugal, on the other hand, the incidence of part-time work is less than per cent (Appendix, table A). Occupational segregation by hours of work in our sample of countries ranges from 30 per cent in Ireland to 52 per cent in Portugal, with an average of 40 per cent. Except for Belgium, Portugal and Spain, segregation is greater for men than for women in all countries (figure 1). The intuitive explanation for greater segregation for men is that women are often confined to a narrower range of occupations, wherefore segregation due to part-time work appears to be more restrictive for men (Sparreboom, 2014). The pattern by sex is the same for youth, and segregation for young women exceeds that for young men only in Belgium (figure 2). At the same time, segregation for youth is greater than segregation for adults in all countries, except for Greece, Italy, Portugal and Spain. This seems to be consistent with the greater role played by part-time work in conjunction with other activities for young workers in comparison with adults. Figure 1. Segregation by hours of work (ID, percentages), by sex, Ireland Spain Finland Belgium Austria Sweden United Kingdom France Average Luxembourg Italy Greece Denmark Netherlands Norway Portugal All Male Female Source: Author s calculations based on the Eurostat ELFS.

7 Occupational segregation by hours of work in Europe 71 Figure 2. Youth segregation by hours of work (ID, percentages), by sex, Greece Spain Finland Sweden Italy Average Denmark Ireland France Austria United Kingdom Belgium Netherlands Luxembourg Portugal Norway All Male Female Source: Author s calculations based on the Eurostat ELFS. We were also interested in finding out which subgroups of the population were driving the overall levels of segregation by hours. This could be determined by examining the pairwise correlations between levels of segregation in subgroups of the employed population. These correlations show that the overall level of segregation is strongly and significantly correlated with the level for adults (who constitute the large majority of the employed population) but not with the level for youth (table 1). The correlation between levels of segregation across all workers and adult male workers is particularly strong, and the same is true across all young workers and young male workers. Table 1. Correlations of levels of segregation by hours of work (ID) between subgroups of the employed population, 2011 All Male Female Adult Adult male Adult female Youth Youth male Youth female All 1.00 Male 0.90*** 1.00 Female 0.75** 0.61* 1.00 Adult 0.87*** 0.66** 0.69** 1.00 Adult male 0.78*** 0.78*** 0.65** 0.82*** 1.00 Adult female 0.67** 0.45* 0.92*** 0.81*** 0.69** 1.00 Youth Youth male 0.55* 0.63* *** 1.00 Youth female ** 0.50* 1.00 Note: * p < 0.1; ** p < 0.01; *** p < Source: Author s calculations based on the Eurostat ELFS for 15 countries.

8 72 International Labour Review There is also a significant correlation between levels of segregation by hours for male and female workers, across all workers and for both adults and youth. This seems to be consistent with Tijdens s (2002) finding that the responsive firms regime is important for explaining the incidence of parttime work, to the extent that firms are equally open to demands for reduced hours from both men and women. In this case, men benefit from the fact that part-time work is already accepted due to the strong presence of women in certain occupations. Segregation for selected subgroups Table 2 shows the levels of segregation by hours for selected subgroups of the employed population. In general, the levels suggest that workers who are often in a disadvantaged labour market position also face more restrictions on occupational choices, although this does not necessarily hold true for all subgroups. Segregation by hours of work is generally lower for workers with a permanent job compared with those who have a temporary job; it is generally lower for those who do not want to work more hours compared with those who are looking for more work; and it is generally lower for those who have a high level of education (defined as higher than secondary education) compared with those who have a low level of education. Across all workers, there is no difference in the levels of segregation between foreign-born and nativeborn workers, but there are large differences between foreign-born adult male workers and young male workers, who appear to face more restrictions than their native-born counterparts. Young female workers with low levels of education are subject to relatively higher levels of segregation, but the opposite is true for the adult male and female subgroups. Table 2. Segregation by hours of work for selected subgroups of the employed population (ID, percentages), 2011 All Male Female Adult Adult male Adult female Youth Youth male Youth female All Permanent job Temporary job Native-born Foreign-born No more work wanted More work wanted High education level Low education level Note: High education level is defined as higher than secondary education. Source: Author s calculations based on the Eurostat ELFS for 15 countries.

9 Occupational segregation by hours of work in Europe 73 Drivers of segregation by hours of work Segregation by hours of work can be expected to be related to three groups of factors: (1) quantity of employment, (2) quality of employment, and (3) institutions. If levels of part-time work are very low, it seems likely that there would be no part-time workers in some occupations, while at higher levels of parttime work, there would be broader coverage of occupations, thereby lowering segregation. Segregation may be similarly affected by overall levels of employment and unemployment. Moreover, segregation is also related to the quality of employment if part-time jobs are associated with low-quality employment and such employment is more likely in certain sectors and/or occupations. Finally, institutions are important as factors that affect part-time employment rates in the long run, and they may also be more important in certain sectors and/or occupations, thus also affecting segregation. The abovementioned three groups of factors clearly give rise to large sets of indicators that potentially help to explain segregation. Table 3 lists a selection of indicators for each group of factors, together with the definitions and data sources. The first group consists of indicators that are widely used to measure the volume of employment, and it includes the part-time employment rate to capture the volume of part-time employment. The indicators in the second group are related to the quality of employment, such as job security, overtime hours, as well as education. Even though the latter indicator refers to the worker (and not the job), it is likely that workers with a higher level of education would enjoy a higher quality of employment. These indicators regarding the quality of employment have been calculated as the differences between the share of jobs or workers with certain characteristics among full-time workers and the corresponding share among part-time workers. We have added the share of part-time workers who want to work more hours as a separate indicator that does not belong to any of the three groups, given that this share may also affect occupational choices. In contrast to the indicators on the quantity and quality of employment, the three institutional indicators (union density, family benefits and marginal tax rates) are not based on the ELFS, and they are only available across all workers. The results in table 4 show that all the indicators are significantly correlated with segregation for at least one subgroup of the employed population, but none of the indicators is correlated with segregation across all groups. In other words, and despite the significant correlations highlighted in table 1 above, the correlations between segregation in subgroups (which are often driven by different factors) are not strong enough to drive results across all workers. The part-time employment rate only correlates (negatively) with segregation for adult men. For all countries, this is the subgroup with the lowest share of workers in part-time employment, therefore it is no surprise that a higher part-time employment rate is associated with lower segregation by hours for this subgroup. Adult men also constitute the largest subgroup of the employed population, and they are often relatively equally distributed across occupations (in full-time employment). For men (of all ages), a higher

10 74 International Labour Review Table 3. List of indicators potentially affecting segregation by hours of work Indicator group Short name Indicator definition Unit Source For each subgroup I PTR Share of part-time work in employment Percentage Eurostat ELFS (part-time employment rate) UNR Unemployment rate Percentage Eurostat ELFS EPR Employment-to-population ratio Percentage Eurostat ELFS II TEMP Difference between the share of workers with temporary contracts among full-time workers and the corresponding share among part-time workers EDUC Difference between the share of workers with a low education level among full-time workers and the corresponding share among part-time workers OVER Difference between the share of workers with paid overtime hours among full-time workers and the corresponding share among part-time workers INVOL Share of part-time workers wanting to work more Across all workers Percentage points Percentage points Percentage points III UNION Trade union density Percentage of employees FAMBEN TAX Public spending on family benefits in cash, services and tax measures Marginal effective tax rates for part-time employees (one-earner married couple with two children) Percentage of GDP Percentage Eurostat ELFS Eurostat ELFS Eurostat ELFS Eurostat ELFS OECD.Stat en/ictwss OECD, Social Expenditure Database (SOCX) expenditure.htm OECD, Benefits and Wages: Statistics benefits-and-wages.htm employment-to-population ratio (EPR) is associated with more segregation, which is due to the fact that more employment primarily means more full-time employment in cross-country comparisons (there is a correlation between the male EPR and the male full-time EPR, where the EPR is calculated with fulltime work only in the numerator). A higher EPR for young men is also correlated with more segregation, and there is also a strong and highly significant negative correlation between segregation and unemployment for this group. The negative correlation with unemployment may seem to be counterintuitive, but it is due to the respective roles of full-time work and part-time work with respect to unemployment. Higher unemployment is associated not only with less full-time work but also with less part-time work for young male workers, which as a whole reduces segregation (figure 3). For the same reason, higher unemployment is also associated with less segregation for young women. In contrast to this, unemployment is

11 Occupational segregation by hours of work in Europe 75 Table 4. Univariate regressions of segregation by hours of work, 2011 All Male Female Coef. p > t R 2 Coef. p > t R 2 Coef. p > t R 2 PTR UNR EPR ** TEMP * EDUC OVER INVOL * UNION FAMBEN TAX Adult Adult male Adult female Coef. p > t R 2 Coef. p > t R 2 Coef. p > t R 2 PTR ** UNR * EPR TEMP ** EDUC OVER INVOL ** UNION 0.176* FAMBEN 4.039** ** TAX * Youth Youth male Youth female Coef. p > t R 2 Coef. p > t R 2 Coef. p > t R 2 PTR UNR 0.439** ** * EPR ** TEMP EDUC * OVER ** INVOL 0.386** ** UNION FAMBEN * TAX Note: * p < 0.1; ** p < primarily associated with less part-time work for adult women, resulting in increased segregation between full-time and part-time work (figure 4). A similar pattern explains the opposite signs in the correlation between involuntary parttime work for adult women on the one hand and for young men on the other. 8 8 More unemployment is associated with less full-time work, less part-time work and a lower share of involuntary part-time work for young men, and it primarily affects part-time work and the share of involuntary part-time work for adult women.

12 76 International Labour Review Figure 3. Segregation, unemployment and employment of young male workers (percentages) Unemployment rate Segregation (ID) Full-time EPR Part-time EPR Note: The x-coefficients of the three regression lines are significant at the 90 per cent confidence level. Source: Author s calculations based on the Eurostat ELFS for 15 countries. Figure 4. Segregation, unemployment and employment of adult female workers (percentages) Unemployment rate Full-time EPR Segregation (ID) Part-time EPR Note: The x-coefficients of the regression lines for segregation and part-time workers are significant at the 90 per cent confidence level. Source: Author s calculations based on the Eurostat ELFS for 15 countries.

13 Occupational segregation by hours of work in Europe 77 The indicators related to the quality of employment (table 4), show that the difference between full-timers and part-timers with respect to the share of temporary contracts is negatively correlated with segregation for adult women. Given that the share of temporary contracts in almost all countries is higher for part-timers, the negative coefficient in table 4 means that a larger difference in absolute terms is associated with more segregation. Such segregation concurs with the role played by part-time work as a secondary labour market, given that temporary work allows for high-turnover jobs. In the case of young females, the indicator regarding the share of workers with a low level of education points in the same direction. This share is also higher among part-time workers in comparison with full-timers in most countries, and an increasing difference in shares (in absolute terms) is associated with more segregation. However, in contrast to temporary work, low levels of education in part-time workers may also reflect the self-selection of workers into part-time jobs. For example, in dual-earner families where the full-time work of one partner is combined with the part-time work of the other, it may be advantageous for the better-educated partner to work full time. The relevance of self-selection is suggested by the absence of a significant correlation between involuntary part-time work and the level of education for young women, while a significant positive correlation is found between involuntary part-time work and temporary work (figure 5). Finally, greater control over working time is often cited as a positive characteristic of part-time work, which is confirmed by a lower Figure 5. Involuntary part-time work, temporary contracts and education of young female workers (percentages) Share of involuntary part-time workers Share of workers with temporary contracts Share of workers with low education Note: The x-coefficient of the regression line for the share of workers with temporary contracts is significant at the 90 per cent confidence level. Source: Author s calculations based on the Eurostat ELFS for 15 countries.

14 78 International Labour Review Figure 6. Segregation, family benefits and employment of adult female workers (percentages) Family benefits (percentage of GDP) Adult female EPR Segregation (ID) Share of involuntary part-time workers Note: The x-coefficients of the three regression lines are significant at the 95 per cent confidence level. Source: Author s calculations based on the Eurostat ELFS for 15 countries. incidence of overtime work among part-timers in comparison with full-timers in most countries. The indicator that measures the difference in the incidence of overtime work is significantly and positively correlated with segregation for young men. This seems, at least partially, to be a reflection of self-selection (similar to levels of education), given that the incidence of overtime work is not correlated with involuntary part-time work for young men. Institutional indicators appear to be primarily important for adult workers, and the correlations are in line with expectations and previous analyses (table 4). Union density is associated with less segregation for adult workers, which could be expected if unions are attempting to level the playing field between full-time and part-time work in terms of working conditions. Segregation for adult women is reduced by marginal effective tax rates, which measure the extent to which taxes and benefits reduce the financial gain from increased hours of work. In other words, a higher marginal effective tax rate makes it less attractive to switch from part time to full time, which reduces segregation between full-time and part-time work. 9 Family benefits also tend to mean lower segregation for adult women, which seems to be partly due to the positive correlation between family benefits and the EPR. Apart from rais- 9 The marginal effective tax rate is positively correlated with the part-time EPR for adult women.

15 Occupational segregation by hours of work in Europe 79 Figure 7. Segregation, family benefits and education of young female workers (percentages) Segregation (ID) Family benefits (percentage of GDP) Full-time/part-time differences in the share of workers with low education Note: The x-coefficients of the regression lines are significant at the 90 per cent confidence level. Source: Author s calculations based on the Eurostat ELFS for 15 countries. ing the volume of employment for adult women, family benefits also appear to reduce the share of involuntary part-time workers and thus expand occupational choices (figure 6). In the case of young women, there is no relationship between family benefits and involuntary part-time work, but such benefits do affect differences in the share of workers with a low level of education. As it was previously noted, a greater difference in this share is associated with segregation for young women (figure 7). Conclusions The purpose of this paper has been to provide a systematic account of occupational segregation between part-time and full-time work in a sample of European countries. Monitoring such segregation is important because it helps to explain the disadvantages of part-time workers. Such disadvantages particularly affect women and young workers and are partly due to the limited occupational choices of part-time workers. On average, segregation by hours The involuntary element in involuntary part-time work refers to the number of hours of work and not to the occupation of workers. Nevertheless, it seems likely that a reduction in the share of involuntary part-time work also involves occupational choices.

16 80 International Labour Review of work in the 15 countries examined in this paper amounted to 40 per cent, which is lower than the average level of segregation by sex in the EU. 11 Parttime work is more restrictive for men than for women in terms of occupational choice, and in most countries segregation is greater for youth than for adults. In general, segregation tends to be more pronounced for groups that are already in a disadvantaged labour market position. Segregation by hours of work may be explained by many factors, including some of the factors that affect segregation by sex, such as differences in human capital, experience, preferences and stereotypes. The analysis in this paper has demonstrated the relationships between segregation and three groups of indicators, which appear to be important for certain groups of workers but not for others. Such variations are most likely a reflection of the socio-economic and labour market positions of these groups, such as the level of part-time employment (the rate for a particular group). Given these differences and the fact that segregation does not have a single-factor explanation, the analysis clearly does not identify causal mechanisms. Correlations between segregation and unemployment, for example, are dependent on additional indicators that may have an effect, which in turn may raise or lower segregation by hours. Opposite effects were found with regard to institutional/policy indicators, which appear to be particularly relevant for women and young workers but do not have a uniform effect on segregation. In order for part-time work to be freely chosen, as suggested by the ILO s Part-Time Work Convention, 1994 (No. 175), it should be available in a broad range of sectors and occupations. The extent to which this is the case can be monitored using segregation by hours, which should be taken into account together with other indicators that capture the quality of employment. Gender differentials in current levels of segregation by hours suggest that a good starting point would be to promote the convergence of male and female part-time employment rates, particularly in countries where female part-time employment rates are relatively high. Such convergence could be expected to reduce gender inequality in terms of both hours of work and segregation by hours of work. References Abhayaratna, Joanna; Andrews, Les; Nuch, Hudan; Podbury, Troy Part time employment: The Australian experience. Staff Working Paper. Melbourne, Australian Government Productivity Commission, pp Andrews, Martyn; Bradley, Steve; Stott, Dave Measuring pre- and post-labour market occupational segregation using careers service data, in Journal of Education and Work, Vol. 17, No. 1 (Mar.), pp Antecol, Heather; Cobb-Clark, Deborah A. 20. Do non-cognitive skills help explain the occupational segregation of young people? IZA Discussion Paper No Bonn, Institute for the Study of Labor. 11 For example, Bettio and Verashchagina (2009) reported that the level of occupational segregation by sex was 51.0 per cent in 2007 (based on a methodology similar to the one used in this paper).

17 Occupational segregation by hours of work in Europe 81 Bardasi, Elena; Gornick, Janet C Working for less? Women s part-time wage penalties across countries, in Feminist Economics, Vol. 14, No. 1 (Jan.), pp Bettio, Francesca; Verashchagina, Alina Gender segregation in the labour market: Root causes, implications and policy responses in the EU. Luxembourg, Publications Office of the European Union. Buddelmeyer, Hielke; Mourre, Gilles; Ward, Melanie Why do Europeans work parttime? A cross-country panel analysis. ECB Working Paper Series, No Frankfurt, European Central Bank. Charles, Maria; Grusky, David B Models for describing the underlying structure of sex segregation, in American Journal of Sociology, Vol. 0, No. 4 (Jan.), pp Equality and Human Rights Commission (EHRC) How fair is Britain? Equality, human rights and good relations in 20: The first triennial review. Manchester. ILO. 2015a. Global Wage Report 2014/15: Wages and income inequality. Geneva b. Non-standard forms of employment. Report presented at the Meeting of Experts on Non-Standard Forms of Employment, Feb. Geneva Global Employment Trends for Youth 2013: A generation at risk. Geneva. Kjeldstad, Randi; Nymoen, Erik H Part-time work and gender: Worker versus job explanations, in International Labour Review, Vol. 151, No. 1 2 (June), pp Manning, Alan; Petrongolo, Barbara The part-time pay penalty for women in Britain, in The Economic Journal, Vol. 118, No. 526 (Feb.), pp. F28 F51. Matteazzi, Eleonora; Pailhé, Ariane; Solaz, Anne Part-time wage penalties for women in prime age: a matter of selection or segregation? Evidence from four European countries, in Industrial and Labor Relations Review, Vol. 67, No. 3 (July), pp ; ; Does part-time employment widen the gender wage gap? Evidence from twelve European countries. ECINEQ Working Paper Verona, Society for the Study of Economic Inequality. Mulekar, Madhuri S.; Knutson, John C.; Champanerkar, Jyoti A How useful are approximations to mean and variance of the index of dissimilarity?, in Computational Statistics & Data Analysis, Vol. 52, No. 4 (Jan.), pp OECD (Organisation for Economic Co-operation and Development). 20. Employment Outlook: Moving beyond the jobs crisis. Paris, pp O Reilly, Jacqueline; Fagan, Colette Part-time prospects: An international comparison of part-time work in Europe, North America and the Pacific Rim. Routledge, London. Sparreboom, Theo Gender equality, part-time work and segregation in Europe, in International Labour Review, Vol. 153, No. 2 (June), pp Tijdens, Kea G Gender roles and labor use strategies: Women s part-time work in the European Union, in Feminist Economics, Vol. 8, No. 1, pp van Bastelaer, Alois; Lemaître, Georges; Marianna, Pascal The definition of part-time work for the purpose of international comparisons. OECD Labour Market and Social Policy Occasional Papers No. 22. Paris. World Bank World Development Report 2012: Gender equality and development. Washington, DC.

18 82 International Labour Review Appendix Table A. Incidence of part-time work as a percent of the employed population in 15 European countries, 2011 All Male Female Adult Adult male Adult female Youth Youth male Youth female Austria Belgium Denmark Finland France Greece Ireland Italy Luxembourg Netherlands Norway Portugal Spain Sweden United Kingdom Source: Author s calculations based on the Eurostat ELFS.