Segregation and part-time workp. 1
______
Female Employment and Family Formation in National Institutional Contexts (FENICs)
Research Group Erlangen-Nuremberg (FAU)
Segregation and part-time work:
Obstacles to adequate employment:
Stephanie Steinmetz and Johann Handl
January 2003
Contact:
Stephanie Steinmetze-mail:
Johann Handl e-mail:
FENICs working group of the
University of Erlangen-Nuremberg
Faculty of Arts I (Philosophy, History and Social Sciences)
Institute of Sociology
Kochstraße 4
D-91054- Erlangen
Contents
1.Introduction: Part-time work - equalization or marginalization ?...... p. 3
2. Theoretical and methodological assumptions...... p. 5
a) Problems of international comparability of working time...... p. 5
b) Multi-dimensionality of sex-specific occupational segregation...... p. 6
c) Measurement of segregation: some methodological considerations...... p. 8
3. Development of female part-time employment – a short overview...... p. 10
4. Part-time work and the aspects of sex segregation...... p. 13
5. Part-time work and the dimension of sex-typing...... p. 16
6. Conclusion...... p. 22
References...... p. 25
1. Introduction: Part-time work - equalization or marginalization ?
Modern societies are affected by various processes of social change, like the increase of formal education, the shift of demographic structures and the structural change of work. Especially the last-mentioned phenomenon is very often discussed as a crucial part of the social modernisation of societies in their course of the transition from an industrial to a service-orientated economy. As a consequence of these ongoing processes, in all member states of the European Union women have been increasingly involved in the labour market over the past few decades. Ultimately, this development is caused by the fact that the formal education of girls has risen, and that women more and more scrutinise their traditional role as housewife and mother critically.
The European Union underlined the importance of an increase in the female labour market participation at the Lisbon Summit in March 2000 where the employment guidelines for 2001 were adopted by the European Union Heads of State and Government. One of the substantial changes to the former guidelines is that, for the first time, a specific overall target to increase women’s employment from 51% today to 60% by 2010 has been included (European Commission 2001; Annual Report 2000).
The general increase in the female labour market participation is accompanied by the fact that during the last thirty years, in most European countries, part-time work has increased more than full-time work.[1] Frequently, a line is drawn between the expansion of part-time employment in the European Union, a growth of the integration of women in the labour force, and pretended improvements in respect of equality between men and women on the labour market. This dual trend prompts many commentators to assume that part-time work is not just a helpful but an essential element in the mobilisation and integration of female labour force.
Optimistic statements as to part-time work merely focus on the positive side of this kind of work which is often described as flexible and family-friendly. However, part-time work also has its drawback. It is usually associated with words like low-skilled, low-paid and precarious or insecure. In other words, the quality of jobs falls in terms of payand working-hours. In this connection, also the ILO report (1992)demonstrates that part-time jobs are worser than full-time jobs in several respects, including skill levels, wage rates, promotion prospects and relations with management.[2]
Consequently, it could be called into doubt that, in general, a higher participation rate of women alone will lead to more equality between the sexes on the labour market. Also the academic literature on this subject agrees that the increasing labour force participation of women led neither to equal opportunities for both sexes nor to their equal integration into the labour market. Moreover, this development begs the question whether the increase of women in the labour market also improves the quality of jobs taken by women.
Against this backdrop, the existing differences in the distribution of women and men within sectors, occupations and jobs, known as the “sex-specific segregation of the labour market”, have been and continue to be a prominent feature of the labour market. Policy debates often depict occupational segregation by sex as an important source of labour market inefficiency and rigidity. It represents exclusion as well as wasteful use of human resources, since many persons who are qualified and appropriate for an occupation may be excluded because of their sex. The attachment of importance to the sex-specific segregation of the labour market can also be justified when considering the link to sex differences in status or prestige and income (Beller, 1982). It has frequently been deemed an important reason for the male-female wage gap and for women’s lower occupational status. Furthermore, it must be pointed out that the polarisation of labour markets by sex affects the view of men and women and that of women on themselves. It is both cause and consequence of gender stereotypes which have a negative effect on educational and training decisions early in life as well as on later decisions about which parent will disrupt his/her career to look after young children. By that, it perpetuates gender inequalities into future generations. All of this leads to unequal opportunities undermining one of the cornerstones of overall equality in society (Melkas 1997, pp. 342). Accordingly, it is not surprising that international organisations identified occupational segregation as the main barrier to women’s full participation in the labour market, and argued for policies that would integrate all occupations.[3]
Empirical studies on this issue frequently note with surprise that occupational segregation still lingers on while the labour-force participation of women and men has become more similar and, in some countries, wage differentials have become smaller (Anker 1998, Charles 1992, Hakim 1996, Jacobs 1989, Reskin 1994). A contribution of particular importance to this research is the well-known book "Gender and Jobs - Sex Segregation of Occupations” written by Anker 1998. He connects the research on divided labour markets with trans-national research and methodologies, and combines several sub- and mainstream notions on the subject. In the framework of his research activities for the International Labour Organisation (ILO), he also concludes that occupational segregation, surprisingly, has not changed fundamentally in the last decades.
In combining these two aspects, which determine possibilities of equal participation of women in the labour market, the following paper raises the question how far working-time patterns and sex-specific occupational segregation are connected In the following section two, a summary of the theoretical and methodological background to this paper will be given. In this context, the problematic of an international definition of part-time work will be discussed first.Subsequently, a new conceptualisation of segregation will be introduced, and the used data-base will be presented. To give a first impression of the development of female part-time participation, section three offers a short overview of the developments which have occurred during the last thirty years. Following the theoretical distinction of the different aspects of sex-specific occupational segregation, the empirical analysis will begin in section four by using the standardised index of dissimilarity and the index of dissimilarity for the different working times. After that, in section five, we concentrate our multivariate analysis on the aspect of sex typing. Therefore we introduce the concept of “female-dominated”, “integrated” and “male-dominated” occupations to analyse with the implementation of multi-logit models the connection of working time, sex, education and sex-typing. In section six, the main results of the analysis are summarised to show the connection between occupational segregation, sex, working time and education.
2. Theoretical and methodological assumptions
The growing concern with “non-traditional” or “atypical” work and aspects of occupational segregation in an international research design encounters various problems concerning the comparability of variables, quality of data, and validity of indices (Anker 1998, p. 1, Charles & Grusky 1995 and 1998, Hantrais 1996).
a) Problems of international comparability of working time
One of the main problems concerning the international comparability between countries in connection with part-time work is its general definition. According to the definition proposed by the ILO (1992, p. 5), part-time work is defined as regular employment in which working time is substantially less than normal. The same idea is expressed in the statement of Convention No. 175 of the International Labour Conference offering the following definition: “…the term part-time worker means an employed person whose normal hours of work are less thanthose ofcomparable full-time workers”.[4] The definition contained in the European Framework Agreement is very similar to that of the Convention: “The term “part-time worker” refers to an employee whose normal hours of work, calculated on a weekly basis or on average over a period of employment of up to one year, are less than the normal hours of work of a comparable full-time worker”. (European Communities (1998), p.13).
In this context, the problem is obvious: part-time work is defined in relation to full-time employment. The number of hours worked by a full-time worker in a certain industry or occupation serves as reference point. The number of working hours in full-time jobs, however, varies not only from activity to activity but also from country to country.
As a consequence, the situation becomes far more complicated if the incidence of part-time work is sought to be compared internationally. The phrase “whose normal hours of work are less than those of comparable full-time workers”, which appears in the ILO Convention, is interpreted differently according to each country’s national law. The same applies to statistical definitions. Some countries draw the line between part-time and full-time work by reference to the usual hours of work per week: 30 hours in Finland, 35 in Austria and Sweden etc. Lastly, some countries, including Germany, the Netherlands, Spain and the United Kingdom, combine the two methods (International Labour Review, Vo. 136, No. 4 1997). For the purpose of this paper, a three-step division is presented in the table below by following the OECD data (Bastelaer, Lemaître & Marianna 1997) but limited through the available data-set.
Table 1: Definition of different working times
Categories / Working hoursMarginal-time / <17 h per week/ data set category 0-17
Part-time / 18-32 h per week/ data set categories 18-27 and 28-32
Full-time / >33 h per week/ data set categories 33-open
Literature concerning flexible working times and women’s reasons for working part-time, most often, brings into focus employees who work between 1 and 30 h per week.[5] However, it is indispensable to differentiate in this regard between marginal and part-time workers. Huge differences in social security and social problems necessitate this differentiation.
b) Multi-dimensionality of sex-specific occupational segregation
As mentioned above differences in the distribution of women and men within sectors, occupations and jobs have been and continue to be a prominent feature of the labour market.
The growing concern with aspects of occupational segregation in an international research design encounters various problems concerning the quality of data, comparability of variablesand validity of indices (Anker 1998, p. 1, Charles & Grusky 1995). There has been a long ongoing debate on the question which single-number index may be the best one for measuring segregation especially for the purpose of international and/or historical comparison. In our opinion this debate suffers from the fact that the concept of “segregation” has rarely been specified exactly. Thus, different indices have been proposed by several authors over the last three decades which all claim to measure segregation, but with varyingconnotations. To clarify the situation, we propose the following distinction between different aspects of segregation:
Some authors restrict their understanding of the term segregation to the "grade of the feminisation" of an occupation or sector. We call this perspective "sex typing". The best - or at least a well known - way to measure this aspect of the labour market is to employ the "Standardised Index of Dissimilarity" (Dst). It varies between 0, where we cannot find a differentiation of occupations as “typical” male or female, up to 1 (100%), where a strict attribution of occupations as “typical” male or female exists.
A further dimension of "segregation" can be found by considering not only the amount of "sex typing" of occupations but also the weight of each occupational category resulting from its size (e.g. number of persons employed). To perform this weighing process, the percentage of males and females working in a specific occupation or sector must be calculated. By summing up the differences in the percentages between males and females the "Index of Dissimilarity" (D) is calculated. It equals 0 in the case of complete equality (where women and men are distributed identically across occupations) and 1 (100%) in the case of complete dissimilarity (where women and men work in totally different occupational groups). We call this aspect "structure of sex-specific chances" and keep "segregation" as a generic term which covers different aspects of sex-specific differences.
At last, we restrict the term "sex-specific occupational inequalities" to a situation in which the perspective of sex-specific chances of accessing different positions is extended so that, in consequence, it is possible to determine the order these positions and the distances between them. For example, this ranking could be aligned with the mean income in occupations or the socio-economic status or prestige. If it is possible to expand the analysis in this way, we can go beyond the level of more or less complex indices of segregation and deploy the powerful tools of metric statistical analysis, i.e. we can calculate “Arithmetic Mean” and “Variance” of a distribution. The following table sums up these considerations in a very short form:
Table 2: Theoretical dimensions of segregation
Theoretical Concept / Measured by / Used Classification“Segregation" : generic term which includes the aspects of
Sex typing of occupations / Standardised Index of Dissimilarity (DST) / ISCO88 3-digit
Sex specific occupational chances / Index of Dissimilarity (D) / ISCO88 3-digit
Sex specific inequalities / Differences in Mean, Variance etc. / ISEI 3-digit
c). The measurement of segregation: Some methodological considerations
Empirical findings on the patterns of segregation are - among others - strongly influenced by the classification utilised for measurement. Typically, some sort of occupational or sectoral classification of the employees constitutes the backbone of all segregation research. With the ISCO88 classification on the 3 digit level included in the data sets, which have been made available to us for research by Eurostat, we are able to distinguish between approximately 116 different occupations. The fundamental logic of this classification is mainly derived from skill requirements, meaning that a ‘lower’ code implicates higher skill, which is defined as “the ability to carry out the tasks and duties of a particular job” (ILO 1990, p. 2). Irrespective of controversy opinions expressed in literature, our own studies on the dependency of results from employing a more or less differentiated occupational classification convinced us to use the 3 digit level as far as possible.[6]. Although the international structure of the ISCO88 classification places some problems concerning the comparability of occupations/years between different countries, it is a very powerful tool not only for labour market and segregation research, but also for stratification analyses. It should be mentioned that the used data base (SFE4)[7] is grounded on an extract of the European Labour Force Survey (ELFS). It was designated for analysing the family situation of women and the number of hours they are working.[8]In this paper, the central variable is “hww” which includes information about the working time. In addition, the variables "country", "sex", “education” and the ISCO88 variable are taken into account.
The emergence of partly contradictory and often surprising results in international segregation research is not only an outcome of the use of different classifications and data but also arises from the use of different methods of measurement which provoked a long and ongoing debate about the way in which an index of segregation should be constructed. Thereby, various indices and log-linear techniques have been proposed by several researchers[9] which all claim to measure “pure” segregation, as the marginal dependencies of traditional segregation-indices (i.e. the Index of Dissimilarity D) have become the main source of criticism. However, as argued above, a theoretical clarification of the concept of “segregation” might overcome the long lasting search for the “best” single-number index. Hence, following our analytical differentiation of three dimensions of “segregation” (i.e. sex-typing of occupations, occupational chances and sex-specific inequalities) we inspected the correlation between various segregation-indices which have been classified according to the conceptual schemesmentioned above. Four indices have been considered for this purpose. The first two are the Index of Dissimilarity D and its standardised version Dst,(Duncan & Duncan 1955) which have already been explained above. The third one is the IP-Index proposed by Karmel & McLachlan (1988) which can be interpreted as the proportion of the workforce (persons in employment) which would have to change jobs in order to remove segregation. Finally, the Index L which is based on log-linear modelling has been developed by ourselves for this purpose (for technical details see Steinmetz, Hedler, Handl 2002; Workingpaper II). Figure1. illustrates the findings on basis of the ISCO88 3-digit classification using ELFS-data of all 15 EU Member States.
Figure 1: Correlation between different measures of segregation in 2000: ISCO88 3-digit
Measures of “sex-typing”
Measures of “occupational chances”