Are all “female-dominated occupations” poorly paid?
Anja Hall, Ana Santiago Vela
Occupations with a very high proportion of women are designated as female-dominated occupations. Wages in female-dominated occupations are on average lower than in typical male-dominated occupations and in mixed occupations not dominated by women. Whether this average income-reducing effect of female-dominated occupations differs across educational levels and across occupational fields has not yet been analysed and forms the focus of the present article. Using the 2018 BIBB/BAuA Employment Survey, we first investigate whether the effect of female-dominated occupations is different for the group of employed persons with a vocational education and training (VET) degree than for the whole group of employed persons with all educational levels. Second, by focusing on employed persons with a VET degree, we analyse whether all female occupations are poorly paid or whether any differences arise depending on the specific field of the female-dominated occupation.
Occupational gender segregation and wages
The German labour market is characterised by a high degree of occupational sex segregation which has scarcely reduced over recent decades (cf. HAUSMANN/KLEINERT 2014; BUSCH 2013 a). Occupations typically held by women are found in areas such as nursing, education, and cleaning, whereas technical and manufacturing occupations are typical male domains. This division is problematic if it is systematically linked with inequalities between women and men. The fact that wages are lower in occupations with a high proportion of women may go some way towards explaining why women on the German labour market earn less than men on average.
The relationship between the proportion of women in an occupation and wage disadvantages is well documented. The average wage in occupations with a high proportion of women is below the average wage in male or mixed occupations (cf. ACHATZ 2018; BUSCH 2013 a). The results of HAUSMANN/KLEINERT/LEUZE (2015) also show that a rising proportion of women in an occupation leads to a drop in the occupation-specific wage level. The authors argue that this is much likelier due to the allocation of women to certain worse paid tasks during the emergence of the modern occupational structure rather than a consequence of a general devaluation of female-dominated occupations.
The question as to why income varies in line with the gender composition of an occupation has been frequently examined, but no conclusive answer has as yet been found. Numerous mechanisms have been analysed. Examples include lower levels of normative overtime in typical female-dominated occupations or extent of part-time work (cf. BUSCH 2013 a; LEUZE/ STRAUß 2009, 2016), the degree of trade union organisation or the existence of binding collective wage agreements (cf. BERNINGER/ SCHRÖDER 2017; LEUZE/STRAUß 2009), differences in occupational specificity (cf. MURPHY/OESCH 2016), and the cultural devaluation of tasks typically performed by women (cf. LIEBESKIND 2004; BUSCH 2013 b). However, the empirical findings relating to the cultural devaluation of tasks typically performed by women are not uniform. The analyses of BUSCH (2013 b) cannot confirm he thesis of devaluation of female-typical tasks for women. Moreover, LIEBESKIND (2004) shows that only certain female-typical tasks such as typing, cleaning and retail work reduce income, whereas tasks like nursing and education do not (cf. also LEUZE/STRAUß 2016 for women with university degrees).
A differentiated look at occupations with a high proportion of women
Previous results at least suggest that tasks in occupations with a high proportion of women do not exert a negative impact on income and also do not have this effect across the board. This finding indicates heterogeneity within female-dominated occupations. Characterisation of a certain occupation as a female-dominated occupation relates exclusively to the proportion of women in the occupation. In this sense, occupations which may differ vastly in terms of occupation-specific fields are all categorised as female-dominated occupations. A differentiated look at female-dominated occupations depending on the specific occupational field may therefore be instructive. This article considers the average negative effect of female-dominated occupations on income differentiating by educational level and by occupational field. This differentiation seems to be useful, because employed persons with VET degrees usually perform tasks in different fields comparedto the tasks usually carried out by persons with academic degrees. In addition, the gender distribution appears to be more balanced in highly qualified occupations (cf. ACHATZ 2018). This means that female-dominated occupations are mostly localised below the academic requirement level. The occupationally differentiated consideration of female occupations in the article is restricted to employed persons who have completed VET.
Database and empirical approach
The 2018 BIBB/BAuA Employment Survey is a current representative survey of around 20,000 workers in Germany (cf. Information Box 1).
Analyses relate to 18,743 employed persons aged between 20 and 65. They include 9,083 employed persons whose highest educational level is a VET degree (4,851 women and 4,232 men). The microcensus surveys from the years 2016, 2017 and 2018 were additionally used to identify the proportion of women in the occupations. All occupational categories contained within the German Classification of Occupations (KldB 2010, 5th digit of the occupation exercised) with a proportion of women of 80 percent or more are considered to be female-dominated occupations.1 The comparison group consists of male-dominated occupations (proportion of women of 20% or less) and mixed occupations (proportion of women more than 20% and less than 80%) . Around 20 percent of the employed persons exercise a female occupation (37% of the women and 4% of the men). Considering employed persons with a completed VET degree only, the proportion of persons working in female-dominated occupations is slightly higher (24 percent – made up of 47% of the women and 4% of the men). For this educational group, female-dominated occupations were further differentiated by occupational fields in cleaning services, sales occupations in the retail trade, office clerks and secretaries, accounting and administration, occupations in healthcare and nursing, doctors’ receptionists and assistants, occupations in geriatric care, teaching and education and other female occupations (for details of allocation of the female-dominated occupational categories [5-digit codes of the KldB 2010] to the groups, please see the Electronic Supplement to accompany this article; cf. information provided at the end of the article).1
2018 BIBB/BAuA Employment Survey (ETB 2018)
The data was collected by Kantar Public of Munich during the period from 2 October 2017 to 5 April 2018 using computer-assisted telephone interviews (CATI). The selection of telephone numbers was based on a random mathematical and statistical procedure (Gabler-Häder sampling process). This ensured that the sample is representative. Landline numbers were contacted, and 30 percent of calls were made to mobile numbers (so-called dual frame approach). The statistical population comprises employed persons aged 15 and above (not including trainees) who work at least ten hours per week (“gainfully employed persons”). Data was adapted to the structures of the statistical population via weighting in accordance with central characteristics on the basis of the 2017 Microcensus. Further information on the BIBB/BAuA Employment Surveys is available on the project site (www.bibb.de/arbeit-im-wandel).
We estimated multivariate models controlling for several characteristics to avoid bias in the income effects. The characteristic of gender, for example, needs to be controlled for because the proportion of women in female-dominated occupations is very high by definition and women, on average, earn less than men. The estimations are based on linear multi-level models in order to account for differences between the occupations (level 1: employed persons, level 2: occupations). The dependent variable is the logarithmised gross hourly wage of the employed persons (cf. Information Box 2). One benefit of this transformation is that the regression coefficients can be approximately interpreted as a percentage change to the average gross hourly wage of a group (in this case female-dominated occupations) as compared with the average gross hourly wage of a reference group (in this case male-dominated/mixed occupations).
Empirical approach – multi-level models (random effect models) take into consideration the hierarchical data structure and are able to map the dependencies within and between the level of individuals and of occupations (defined according to the occupational categories of the KldB 2010) (cf. Hox 2010, p. 211). Multi-level models thus allow the explanatory effect of individual or occupational characteristics for wages to be depicted separately. On the basis of the ETB 2018, around 38 percent of the average gross hourly wage can thus be explained via exercised occupations (i.e. via differences in the occupations performed) (intraclass correlation).
Operationalisation of income – the question aimed at measuring income was formulated as follows. “Now we come to your gross monthly earnings. This means wages or salary before deductions for tax and social insurance. Please do not include child benefit. What is your gross monthly income from your job as a <display job>?” Missing income information was imputed. Gross hourly wage was calculated on the basis of gross monthly earnings divided by hours worked per month (weekly working time *4.35). In cases where overtime was remunerated via time off in lieu, actual working time was replaced by agreed working time.
Less income in female-dominated occupations
Figure 1 presents how female-dominated occupations and wages are related. The male-dominated and mixed occupations act as the comparison group. Shown are the effects of female-dominated occupations on the (logarithmised) gross wage, which are to be approximately interpreted as a percentage deviation compared to male-dominated or mixed occupations. Model M 1 does not initially contain any control variables and represents the bivariate correlation between a female-dominated occupation and income. It is revealed that the wage of employed persons who exercise female-dominated occupations is 24 percent lower on average than the wage of persons employed in male-dominated or mixed occupations. Model M 2 controls for the characteristics of gender, occupational experience, migration background, federal state, age, and self-employment. The income-diminishing effect of a female-dominated occupation is reduced to - 0.17, i.e. female occupations are associated with a gross hourly wage that is 17 percent lower in Model M 2. The primary explanation for the decrease in the income-diminishing effect is the lower wage earned by women, who form the majority of workers in these occupations (“gender pay gap”). If we additionally control for the educational level of employed persons in Model M 3, the negative effect of a female-dominated occupation is further reduced, so that those employed in female-dominated occupations earn a gross hourly wage that is 14 percent lower on average. The explanation here is that the proportion of persons with an academic degree in female-dominated occupations (15 percent) is significantly lower than in male-dominated or mixed occupations (31%) and those holding an academic degree achieve a higher average income than persons with VET degrees.
If only persons with a VET degree are considered (cf. Remark on M 4 in Figure 1), then the effect of a female-dominated occupation for those persons with a VET degree is less negative than for the whole group of persons with all levels of education. Workers with VET thus achieve wages which are only an average of nine percent lower if they work in a female-dominated occupation (as compared to employed persons with a VET qualification who work in male-dominated/mixed occupations). Exercising a female-dominated occupation has a significantly stronger negative effect for persons in possession of an academic degree (not shown in the graphic). This relates to the fact that female-dominated occupations are less likely than male-dominated and mixed occupations to be found at an academic level (5% as opposed to 26%). For this reason, persons with an academic degree who are exercising a female-dominated occupation are more likely to be overqualified for the task they are performing.
Less income in female-dominated occupations – does this apply to all occupations?
The starting point for further occupational analyses is the average negative effect of female-dominated occupations among employed persons with a VET degree in Model M 4 (-0.09) and the question whether the effect remains the same across different fields. The results in Figure 2 investigate this issue and show that two of the nine groups considered (occupations in cleaning services and sales occupations in the retail trade) are associated with a significant strong income-diminishing effect. Employed persons with a VET degree working in female-dominated cleaning occupations achieve a gross hourly wage which is 40 percent lower on average than the wage earned by workers with VET degrees who are in male-dominated or mixed occupations. In female-dominated occupations in retail services, average gross hourly wage is 25 percent less than in male or mixed occupations. A significantly lower income (-0.11) is also achieved in the category of other female-dominated occupations. Female occupations in the categories of doctors’ receptionists and assistants and geriatric care and female-dominated occupations in education are also associated with a lower income, although the effects are not statistically significant in this case (the confidence intervals cross the zero line). It is also possible to recognise that some occupations even tend to be associated with wage benefits. This is the case for those groups which exhibit positive coefficients for the effect of a female-dominated occupation. The wage differences in these cases are, however, not statistically significant either. Nevertheless, workers in female-dominated occupations in the fields of office clerk and secretary, in the field of accounting and administration and in the field of healthcare and nursing tend to display gross average wages which are comparable with those achieved by employed in male-dominated or mixed occupations. Mention should be made at this point that the incomes shown in the female-dominated occupations do not permit any assertions regarding payment according to performance or to job strain ). Especially in nursing occupations, employees are more likely than in other occupations to be dissatisfied with the income they achieve (cf. HALL et al. 2021).
Are female-dominated occupations generally worse paid?
The differentiated consideration of female-dominated occupations by educational level and by fields compared to male-dominated and mixed occupations has revealed how heterogeneous female-dominated occupations are with regard to income chances. First, the average income-diminishing effect of female-dominated occupations for employed persons with a VET degree is lower than for the whole group of employed persons with all educational levels. Second, the results for the employed persons with VET degrees suggest that the (monetary) value of female-dominated occupations is not systematically lower across all fields, so that lower wages than that received in male-dominated and mixed occupations does not occur in all female-dominated occupations. Future research should investigate the heterogeneity within female-dominated occupations even more closely. This could also deliver an important contribution for policies and for career choice to the extent that differentiated analysis of occupations and of the income and employment conditions associated with them would provide a means of relativising a blanket negative connotation of occupations with very high proportions of women. Further research requirements exist with regard to non-monetary employment conditions in female-dominated occupations, whose improvement is particularly needed in the area of nursing (cf. HALL et al. 2021). A look at occupations typically exercised by men would also prove helpful to shed more light. These occupations could likewise be very heterogeneous in respect of income and employment conditions. From the perspective of vocational guidance, it thus remains important to undertake a differentiated consideration of the specific aspects of occupations in a way that goes beyond aggregating characteristics.
A full list of female-dominated occupations (occupational categories from the KldB 2010 with a proportion of women of 80 percent or above) and details of group allocation are included as an electronic supplement and may be accessed at www.bwp-zeitschrift.de/en/bwp.php/en/bwp/grafik/543.
ACHATZ, J.: Berufliche Geschlechtersegregation. In: ABRAHAM, M; HINZ, T. (Ed.): Arbeitsmarktsoziologie. Berlin 2018, pp. 389–435
BERNINGER, I.; SCHRÖDER, T.: Inklusion oder Schließung? Gewerkschaftlicher Organisationsgrad, berufliche Geschlechtersegregation und der Gender Pay Gap. In: Zeitschrift für Arbeit, Organisation und Management 24 (2017) 2, pp. 174–195
BUSCH, A.: Die berufliche Geschlechtersegregation in Deutschland. Wiesbaden 2013 a
BUSCH, A.: Der Einfluss der beruflichen Geschlechtersegregation auf den "Gender Pay Gap". In: Kölner Zeitschrift für Soziologie und Sozialpsychologie 65 (2013 b) 2, pp. 301–338
HALL, A. u.a.: Ansehen und Beschäftigungsbedingungen in der Kranken- und Altenpflege: Stimmen die Berufsbilder in der Bevölkerung mit der Realität überein? Version 1.0. Bonn 2021
HAUSMANN, A.-C.; KLEINERT, C.: Berufliche Segregation auf dem Arbeitsmarkt: Männer-und Frauendomänen kaum verändert (IAB-Kurzbericht Nr. 9). Nürnberg 2014
HAUSMANN, A.-C.; KLEINERT, C.; LEUZE, K.: Entwertung von Frauenberufen oder Entwertung von Frauen im Beruf? In: Kölner Zeitschrift für Soziologie und Sozialpsychologie 67 (2015) 2, pp. 217–242
HOX, J. J.: Multilevel Analysis. Techniques and Applications. 2. Aufl. New York 2010
LEUZE, K.; STRAUß, S.: Lohnungleichheiten zwischen Akademikerinnen und Akademikern: Der Einfluss von fachlicher Spezialisierung, frauendominierten Fächern und beruflicher Segregation. In: Zeitschrift für Soziologie 38 (2009) 4, pp. 262–281
LEUZE, K.; STRAUß, S.: Why do occupations dominated by women pay less? How ‘female-typical’ work tasks and working-time arrangements affect the gender wage gap among higher education graduates. In: Work, Employment and Society 30 (2016) 5, pp. 802–820
LIEBESKIND, U.: Arbeitsmarktsegregation und Einkommen. Vom Wert "weiblicher" Arbeit. In: Kölner Zeitschrift für Soziologie und Sozialpsychologie 56 (2004) 4, pp. 630–652
MURPHY, E.; OESCH, D.: The feminization of occupations and change in wages: A panel analysis of Britain, Germany, and Switzerland. In: Social Forces 94 (2016) 3, pp. 1221–1255
DR. ANJA HALL
Academic researcher at BIBB
ANA SANTIAGO VELA
Academic researcher at BIBB
Translation from the German original (published in BWP 4/2021): Martin Kelsey, GlobalSprachTeam, Berlin