Can work value orientations explain the gender wage gap in Austria?

Abstract

This study analyzes whether work value orientations can explain gender wage differences in Austria. It initially assumes that women have more intrinsic work value orientations than men and are, therefore, more willing to accept low incomes. By contrast, men share more extrinsic work values and are less willing to accept low wages. These assumptions were analyzed empirically with the Social Survey Austria 2016. No or only a weak effect can be found for extrinsic work value orientations. Intrinsic work value orientations have a significant, but opposite effect Higher intrinsic work value orientations – ceteris paribus – result in higher incomes. This lowers the gender wage gap, because women report higher intrinsic work value orientations.

Keywords:

• Gender
• wage gap
• work value orientations
• value acquisition

1. Introduction

Income differences by gender¹ are well-documented in most countries. The OECD (2021) data on gender wage gap of its 29 member countries show Romania, Luxembourg, Belgium, Bulgaria, and Greece having narrow gaps, with values below 5%, whereas Korea, Estonia, Japan, Israel, and Latvia having wide gaps, with values above 20%. Gender differences in wages are discussed in the public and political spheres at regular intervals, especially on specific occasions, such as the International Women’s Day (March 8) and Equal Income Day (Europe, November 20; European Commission 2020).

In Austria, the OECD (2021) has reported a 15.4-percent difference between the median gross earnings of full-time male and female employees while Eurostat (2021) has recorded a 19.9-percent difference in average hourly gross earnings in 2019. Eurostat (2021) reports higher values of gender pay gap only for Estonia and Latvia. Hence, Austria has a high gender wage gap. This finding can be explained by the fact that Austria is a conservative-corporatist welfare state, where women remain responsible for care and household despite an increase in their labor market participation. Austria’s labor market and educational systems are highly segregated by gender, which contributes further to reinforcement of gender stereotypes and roles.

Böheim, Fink and Zulehner (2021) analyzed the gender gap in Austria by applying the Blinder-Oaxaca decomposition (Blinder 1973; Oaxaca 1973), which splits the variables explaining wage discrimination between gender into two components²: one covers occupational and social variables, such as age, working hours, work experience, occupation, occupational position, economic branch, and educational level; the other includes residual, non-measured factors. Their analysis showed that about 34% of the wage difference in 2015 remained unexplained. One reason for the high residual variance could be that the authors concentrate on “objective” factors that are associated with human capital, occupation, economic branch, working hour, job career, etc.

Economists have started including other factors, such as risk aversion, competencies, and gender identity, in their analysis of gender pay gap (Blau and Kahn 2017) but still largely ignored value orientations and attitudes. From a sociological viewpoint, this is astonishing because values have been seen as an important determinant of behavior since the pioneering work of Weber (1980 [1922]) and other founding fathers of sociology. This assumption of the pivotal importance of values can be combined with the conjecture that women prioritize other things over earning a high income. In other words, women are willing to accept a lower income if other values are met. Studies on work value orientations (see Section 2.2) have indicated that women may favor intrinsic over extrinsic work values. However, some of these studies have mainly focused on gender differences in work value orientations, whereas the others have concentrated on the effects of work value orientations on different psychological or behavioral outcomes, such as job satisfaction, health, well-being, and work engagement. Studies analyzing the relation between work value orientations and income are available in the United States (Johnson 2001; Johnson and Monserud 2012; Johnson and Mortimer 2011; Mortimer and Lorence 1979). However, there is no study that examines thoroughly the gender differences in work value orientations and the latter’s link to the gender wage gap. The present study addresses such a research gap.

The remainder of this paper is organized as follows. Section 2 discusses the issue from a theoretical perspective and includes empirical research on work value orientations. Section 3 provides an overview of the data and the methods used. Section 4 presents the results, which are summarized and discussed in Section 5.

2. Theoretical background

2.1. Work value orientations

Work value orientations can be defined as “work-related reinforcement preferences, or tendencies to value specific types of incentives in the work environment” (Malka and Chatman 2003:738). They are seen as individual preferences, wishes, and expectations of certain work values. The label “value” is used because the orientations are general: they do not refer to a particular content or a specific occupation, such as doctor, lawyer, and carpenter, but can apply to all occupations. People differ in their individual evaluation of these values, but the values alone are a part of a society’s culture and linked to basic human values. For example, some authors (Shevchuk, Strebkov and Davis 2018; Vansteenkiste et al. 2007) draw on self-determination theory (Ryan and Deci 2000) and refer to feeling autonomous, related, and competent as basic human needs that underlie work value orientations. With reference to general needs, they agree with other value theories, such as that of Schwartz (2012) who defines human values in relation to basic human needs. However, work values are less universal than other values because they focus on a certain sphere of life. Therefore, some authors define work values as attitudes (e.g., Goldthorpe as quoted in James 2015). From a psychological perspective, work value orientations can be viewed as personal resources (van den Broeck et al. 2011).

A frequently used classification of work values differentiates between intrinsic and extrinsic values (Shevchuk et al. 2018; Vansteenkiste et al. 2007). Intrinsic work values focus on the content of the work. Vansteenkiste et al. (2007:253) “[…] consider[s] an intrinsic work value orientation to reflect employees’ natural desire to actualize, develop and grow at the workplace (i.e., self-development), to build meaningful and satisfying relationships with colleagues (i.e., affiliation) and to help people in need (i.e., community contribution).” By contrast, extrinsic work values refer to work as a means to achieve prestige, status, and high income.

Some authors have differentiated further this basic classification of work values. For example, Pitacho, Palma and Correia (2019) quote the work of Bellah (1996), who differentiates work value orientations into job, career, and calling. The first two orientations represent extrinsic work values: persons with a job orientation focus on materialist aspects, such as high income, whereas those with a career orientation are motivated by upward mobility and high prestige. In his seminal paper, Kalleberg (1977) differentiates four extrinsic values (i.e., convenience, finance, relationships with coworkers, and career). Other authors define the subdimensions of intrinsic values and refer to the importance of social or altruistic work values (Busch 2013; Pollmann-Schult 2009). Still others use a multidimensional concept that sees other values at the same level of abstraction as intrinsic and extrinsic values (Johnson and Monserud 2012; Zou 2015). Even though there may be different concepts, almost all include this fundamental distinction between intrinsic and extrinsic work values. Therefore, we focus on this distinction in the following theoretical part of the study and refer to the sub-dimensions only if necessary. Whether or not further differentiation is needed will be examined empirically.

2.2. Gender differences in work value orientations

Theories of value orientations (e.g. Inglehart 1977; Schwartz 2012) assume that these orientations are the products of socialization during the formative years. While value orientations are considered relatively stable, critical life events or changes in living conditions, such as starting a family, cases of illness in the family, and job loss, can necessitate an adjustment of existing values. With regard to this general model of value acquisition, Rowe and Snizek (1995) distinguish between a gender socialization model and a social structural model to explain the acquisition of work values. The social structural model focuses on the actual social characteristics of the job and the other spheres of life, especially the family, and maps the above-mentioned life circumstances and events that determine a person’s placement within the social structure. In another paper, Vaus and McAllister (1991) differentiate between a job, a family role, and a social characteristics model.

Although Rowe and Snizek (1995) and Vaus and McAllister (1991) were published three decades ago, they are still frequently cited. Their reviews of the existing literature indicate mixed findings, and their own studies result in no or only minor gender differences. We reviewed recent studies with due care³ (Bayrakova 2019; Brožová 2019; Busch 2013; Clark 1997; Doorewaard, Hendrickx and Verschuren 2004; Frankel et al. 2006; Johnson and Mortimer 2011; Miles 2013; Pollmann-Schult 2009; Sharabi, Simonovich and Shahor 2019; Shevchuk et al. 2018) and found that they indicate a convergence in work orientations of men and women on the one hand. On the other hand, intrinsic work values, especially altruistic or social work values, seem still to be more important for women than men, at least in some countries and/or in combination with other factors (e.g. Doorewaard et al. 2004), such as age and education. Gender differences in extrinsic work value orientations are absent or smaller than those for intrinsic work value orientations.

These different findings can be explained with reference to the general value-acquisition model mentioned above. This model assumes that gender differences can be expected in work values in countries or groups (e.g., social and ethnic) if social structural differences between genders are present and gender-specific socialization exists or at least has existed in the past. If traditional gender-based socialization prevails or has prevailed, women tend to develop intrinsic and men extrinsic work value orientations, because this is in accordance with traditional gender stereotypes and roles. In this tradition, women are expected to be sociable, caring, and communicative and men to be competitive and strive for social and occupational advancement and a high income.

Whether or not these work value orientations that one internalizes during socialization will change over the life course depends on the prevailing gender-based division of labor and the often corresponding structure of the labor market. A traditional gender-based division of labor and a labor market strongly segregated by gender provide a few reasons to change one’s internalized work value orientations. A change in work may be prompted by the occurrence of critical life events, such as divorce, job loss, and permanent problems like ill health and/or financial burdens. Intrinsic values, in the case of financial burdens, or extrinsic values, in the case of illness, are side-lined by new urgent priorities.

Gender-based differences in socialization and social structure are embedded in a societal welfare state regime. Esping-Andersen (1990) differentiates three main types of welfare states: liberal, social democratic, and conservative-corporatist. Austria belongs to the latter category. Conservative-corporatist welfare states pursue social welfare and family policies focusing on traditional family models and values, such as the male breadwinner model. By placing more emphasis on institutions of family policy and the division of labor between state and family, scholars of family and gender studies have elaborated upon Esping-Andersen’s typology (Gauthier 2004; Leitner 2003; Lewis 1992; Ostner 1995). For example, Leitner (2003) takes up Esping-Andersen’s notion of de-familialism and develops it further into a concept in which she distinguishes between different types of familialism and de-familialism. By contrast, Saxonberg (2013) prefers the concept of de-genderization to that of de-familialism. He uses the term de-genderization to characterize political measures aimed at accelerating the elimination of gender roles, whereas genderization characterizes a policy that reinforces or supports existing gender roles.

2.3. Gender, work value orientations, and earned income

Work value orientations can have a direct and indirect effect on earned income. Persons who have been socialized toward intrinsic work values may opt for specific educational tracks and jobs (Busch 2013) that in their opinion correspond to their value orientations. These jobs are frequently linked to lower wages, such as jobs in the healthcare and social sectors, especially if a segregated labor market with income differences exists. Therefore, these decisions generate an indirect negative effect on income via educational and job decisions.

Several studies (Judge et al. 2010; Kalleberg 1977; Malka and Chatman 2003; Sheldon et al. 2010; van den Broeck et al. 2011; Vansteenkiste et al. 2007; Zou 2015) have analyzed the effect of work value orientations on job satisfaction, general life satisfaction, and well-being, as well as work engagement, commitment, and job change. In summary, intrinsic work value orientations are associated with positive job outcomes, whereas extrinsic work value orientations are not or negatively associated with positive outcomes. Therefore, if persons with high or low intrinsic work value orientations have the same job, those with high intrinsic work value orientations may less often ask for a wage increase because they are more satisfied than those with low intrinsic work values, which results in a negative effect on income.

A positive effect of the intrinsic work value orientation on income is possible, too. Because persons with a high intrinsic value orientation are more satisfied and motivated, they may be more productive and stay longer at the same firm, which may be reflected in their income (seniority wage). If this is the case, work value orientations may have a positive effect on earnings, women receive – ceteris paribus – a higher wage, and work value orientations will not be able to explain the gender wage gap, given that women are more intrinsically orientated.

However, studies from the United States (Johnson 2001; Johnson and Monserud 2012) have indicated that intrinsic work values are associated with lower income, whereas extrinsic work values are linked to higher income. The relation between extrinsic work value orientations and income can be explained by the assumption that people with extrinsic work value orientations choose jobs that offer them higher wages and upward mobility. The same group is presumably more ready to change their job if their expectations are not met or they find a post that better suits their aspirations. Accordingly, people with a higher extrinsic work value orientation will presumably receive higher wages. Similar to intrinsic work value orientation, extrinsic work value orientation may result in lower income because of lower satisfaction, lower productivity, and more frequent changes in employment.

2.4. Work value orientations and gender wage gap

Concerning our considerations about gender differences in work value orientations on the one hand and the link between value orientations and income on the other, the following hypotheses may partially explain the gender wage gap in Austria.

H1: Female employees have higher intrinsic work value orientations than male employees.

H2: Intrinsic work value orientations have a negative effect on earning income.

If intrinsic work value orientations explain the gender wage gap, both hypotheses must be confirmed.

Accordingly, if extrinsic work value orientations explain the gender wage gap, the following hypotheses must be confirmed.

H3: Female employees have a greater preference for lower extrinsic work value orientations than male employees.

H4: Extrinsic work value orientations have a positive effect on income.

Whether our hypotheses are correct depends on the country being analyzed, as discussed above. In countries with gender differences in socialization and social structure, we expect the hypotheses to hold and work value orientations to contribute to the explanation of the gender wage gap. As described below, Austria is such a country.

Austria is a conservative-corporatist and family-oriented welfare state; it is still oriented toward traditional gender roles and family structures, as well as a gender-based division of productive and reproductive work (Berghammer and Schmidt 2019; Höllinger 2019; Mauerer and Kroismayr 2021). The change in values toward more egalitarian gender roles that has taken place in Europe in the last decades also occurred in Austria (Aichholzer et al. 2019; Bacher et al. 2019b). Most Austrians no longer affirm the traditional division of labor of gender, according to which women are responsible for housework and care work. However, the actual behavior only rudimentarily reflects this change in values. Women still work considerably more in part-time jobs, earn less on average, are responsible for care work in the family, and interrupt their careers more often than men do (Riederer and Berghammer 2020; Schmidt, Kaindl and Mazal 2020). According to Goldscheider, Bernhardt and Lappegård (2015), Austria is characterized by a neo-traditional division of labor. Women carry out household and care work (traditional model) and participate as a part-time worker in the labor market, which is in contrast to the traditional model. In 2016 for example, 47,4% of women were in part-time employment (Fritsch, Verwiebe and Liebhart 2019). Between 2003 and 2016, women’s participation in the labor market has become important to compensate a loss of individual earnings of household members (Hadler and Klebel 2019). The discrepancy between values and behavior illustrates the phenomenon of institutional inertia, as described by England, Allison and Wu (2007). Basic approaches to de-genderization (e.g., a father’s quota for pecuniary aid for childcare) have yet exerted little influence on behavior.

Both the education system and labor market are substantially segregated by gender. Men and women mostly work in so-called gender typical jobs, that is, jobs with a high proportion of one gender (Fritsch 2018; Leitner 2001). Even the education system is already segregated by gender. This facilitates gender-based segregation and prepares for a labor market segregated by gender.

Apart from critical events or changes in the life cycle, this early and pronounced gender-based segregation leads to the fact that, subsequently, no or only smaller work-related adjustments of value orientations occur and that, rather, those work values responsible for the occupational choice in the first place are reinforced. Adjustments in adolescents’ work values after entering the workforce, which consist of, for example, adjusting unreasonably high job expectations to the occupational reality observed in the US context (Johnson 2001; Johnson and Monserud 2012; Johnson and Mortimer 2011), are not expected in Austria to the same extent, because of the earlier educational segregation and the income difference between typical female and male jobs.

Therefore, we assume that the above specified four hypotheses apply in Austria, which means that the wage gap can be explained by work value orientations. It can further be assumed that these differences are more pronounced in older and poorly educated people since they faced stronger gender-based socialization in their youth; attitude surveys provide evidence to this assumption (Berghammer and Schmidt 2019; Höllinger 2019; Mauerer and Kroismayr 2021). To elaborate these and further possible differences in detail is beyond the scope of this paper. However, we will take them into account by testing for heterogeneous effects (see Section 3.3) and in the discussion of limitations (see Section 3.4).

3. Data and methods

3.1. Data

The Social Survey Austria (SSÖ) 2016 (Bacher et al. 2019a) was used to test the hypotheses empirically. The SSÖ is a general social survey comparable to the General Social Survey in the US (http://gss.norc.org/) or Germany (https://www.gesis.org/allbus/allbus). Since its start in 1986, the SSÖ has aimed to analyze social change. To date, five surveys have been conducted, namely, 1986, 1993, 2003, 2016, and 2018. Since the SSÖ 2018 does not contain all the variables necessary for our analysis, we decided to use the SSÖ 2016.

The SSÖ covers the Austrian population aged 16 years and older. A total sample of 2,000 respondents were selected with multistage random sampling; face-to-face interviews were conducted. Our analysis concentrated on employed persons without a migration background (n = 1,115). Persons with a migration background were eliminated because they represent a heterogeneous group. In addition, for those belonging to so-called first-generation immigrants, their socialization has taken place outside of Austria, and our analysis only dealt with socialization within Austria.

3.2. Variables

Work value orientations were measured using eight items. The items were taken from the work value orientation measurement instrument of the corresponding International Social Survey Programme (ISSP) module (GESIS - Leibniz-Institut für Sozialwissenschaften 2017). Table 1 presents the descriptive statistics of the items and results of the exploratory factor analysis. Respondents evaluated having a secure job, an interesting job, and a job that permitted independent working most positively. The lowest valuations were given for the item “good opportunities for advancement.”

Table 1. Descriptive statistics of the items for work value orientations and results of an exploratory factor analysis.

Item	Mean	Std. deviation	Factor 1	Factor 2
Job security (x1)	4.60	0.64	0.21	0.58
High income (x2)	3.85	0.95	−0.22	0.77
Good opportunities for advancement (x3)	3.63	1.14	0.21	0.72
An interesting job (x4)	4.45	0.83	0.70	0.24
A job that allows for independent working (x5)	4.32	0.97	0.70	0.09
A job that allows someone to help other people (x6)	3.85	1.11	0.84	0.04
A job that is useful to society (x7)	4.02	1.08	0.85	0.07
A job that involves personal contact with other people (x8)	4.10	1.11	0.80	0.01
Eigenvalues of rotated factors			3.163	1.516
Explained variance			39.5	19.0
Cronbach’s alpha			0.844	0.478

Source: SSÖ (2016), only employed persons without a migration background, n = 1,115.

Remarks: Scale values range from 1 to 5, with 5 = very important, 4 = important, 3 = neither important nor unimportant, 2 = not important, 1 = not important at all.

An exploratory factor analysis (principal component, varimax rotation with Kaiser’s normalization) revealed two factors. Factor 1 comprises the last five items, x4 to x8, which represent intrinsic values. Factor 2 consists of the first three items, x1 to x3, and can be labeled as extrinsic values. In contrast to other studies (Gesthuizen, Kovarek and Rapp 2019; Pollmann-Schult 2009) that differentiate between intrinsic and social or altruistic work values, social work values are classified as part of intrinsic work value orientations in our data. The reliability of the two factors was acceptable (Taber 2018). Intrinsic and extrinsic work value orientations had Cronbach’s alpha values of 0.821 and 0.480, respectively. Since the reliability of the latter is acceptable, but low, we checked whether it could be improved by the deletion of one item. This was not the case. The deletion of items “job security,” “high income,” and “advancement” lowered the reliability of the extrinsic scale to 0.452, 0.323, and 0.281, respectively. Because the loss in reliability is low if the item “job security” is eliminated, we computed a scale that consists of the two items “advancement” and “high income” and repeated all analyses. We will report whether this narrower operationalization, similar to the meaning of “career orientation” according to Bellah (see Section 2.1), modifies the results.

Gender was used as the main independent variable. The control variables are summarized in Table 2 with reference to the general value acquisition model. The first group contains variables referring to differences in socialization, such as respondents’ age and education and their parents’ socioeconomic status and education. The second group covers life circumstances and critical life events that determine the position of a person in the social structure.

Table 2. Independent and control variables.

Group	Variables	Categories
Independent variable
Gender	Gender	1 = female, 0 = male
Control variables
Other socialization-related variables	Age	1 = 16 to 30 years old; 2 = 31 to 49 years old; 3 = 50 years or older
	Parent’s highest educational level	Education of father or – if higher – education of mother 1 = low; 2 = middle; 3 = high
	Parent’s highest socioeconomic status	Socioeconomic status of father or – if higher – status of mother. Ganzeboom’s international index of socioeconomic status was used (Ganzeboom 2010)
	Respondent’s education	1 = low; 2 = middle; 3 = high
Career-related variables	Interruption of occupation	1 = never; 2 = less than 4 years; 3 = 4 years or longer
Household-related variables	Partnership	0 = no; 1 = yes
	Child 0 to 5 years old	0 = no; 1 = yes
	Child 6 to 17 years old	0 = no; 1 = yes
	Size of place of residence	1 = up to 5,000 inhabitants; 2 = 5,001 to 30,000; 3 = 30,001 to 300,000; 4 = Vienna (approx. 2,000,0000)
Job-related variables	Working time	hours
	Economic branch	12 categories (a)
	Occupation (ISCO)	10 different categories that correspond to the first ISCO level

Note: (a) The 12 branches are: 1 = Production, 2 = Trade, 3 = Agriculture, 4 = Tourism, 5 = Construction, 6 = Communication/Money/Insurance/Housing, 7 = Traffic/Transport, 8 = Professionals/Science/Technical Services, 9 = Public Administration, 10 = Education, 11 = Health and Social Services, 12 = Other Services.

We used gross monthly earned income as the dependent variable, which is reported in income categories by the questionnaire’s respondents. These categories were quantified for analysis using the midpoint of the categories. This conversion results in a gross monthly earned income of EUR 1,806.17 for women and EUR 2,630.76 for men. Therefore, the gender wage gap is 31.3%, which is slightly smaller than the difference of 36.2% reported by Statistics Austria (2021).

3.3. Analysis

Statistical data analysis necessitates the definition of adequate control variables. These variables are not the focus of the study but affect the outcome and correlate with the independent variables that are analyzed. It is important to distinguish between confounders and intervening variables (Bartram 2021). The influence of confounders should be statistically controlled, whereas controlling for intervening variables is problematic and can result in biased estimations.⁴ At times, assigning a variable uniquely to one of the two types is difficult, and forming assumptions about the assignment is necessary (Becker 2005). Controlling for intervening variables may also be useful. They may depend on non-observed (omitted) confounding variables. In this case, not controlling for intervening variables results in a biased estimation of the effect of the independent variable on the dependent variable if the non-observed variables are associated with the independent variables. In cross-sectional data, a third group of potential control variables may be present. The independent and dependent variables may influence each other and are, thus, reciprocally dependent.

We had all three types of control variables in our data. Therefore, we specified different models for data analysis to account for these conditions (see Table 3). In testing H1 and H3, respondents’ age and parents’ highest educational attainment and socioeconomic status could be plausibly interpreted as confounders that influence work value orientations and may be correlated with gender. Therefore, these were controlled variables in Model 1. In Model 1, respondents’ educational level, household, and career-related variables were considered as intervening variables that mediate the effects of gender on work value orientations. However, they may be influenced by non-observed confounding variables; therefore, we also controlled for these variables in Model 2. A reciprocal causal relation exists between job-related variables and work value orientations. Because omitted variables may influence job-related variables, we controlled for job-related variables in Model 3.

Table 3. Models for data analysis.

Testing	Model	Variables of interest	Control variables
H1 and H3	1	Independent variable = gender	Age, parent’s highest educational level and socioeconomic status
		Dependent variable = work value orientations
	2	Independent variable = gender	Model 1 and educational level of respondent, household- and career-related variables
		Dependent variable = work value orientations
	3	Independent variable = gender	Model 2 and job-related variables
		Dependent variable = work value orientations
H2 and H4	4	Independent variables = work value orientations	Gender, age, parent’s highest educational level and socioeconomic status and educational level of respondent, household- and career-related variables
		Dependent variable = income
	5	Independent variables = work value orientations	Model 4 and job-related variables
		Dependent variable = income
H1 and H3	3*	Independent variable = gender and interactions of gender with all other control variables	Model 3
		Dependent variable = work value orientations
H2 and H4	5*	Independent variables = work value orientations and interactions of gender with work value orientations and all other control variables	Model 5
		Dependent variable = income

Models 1 to 3 differed in the restrictiveness of their assumptions. Model 3 was the most restrictive, whereas Model 1 included only control variables that must be controlled for. If Model 3 and all previous models display significant gender effects on work value orientations, H1 and H3 are strongly supported. By contrast, if Model 1 and the following leads to insignificant results, H1 and H3 must be rejected. In the case of mixed findings, H1 and H3 have only weak (two confirmations out of three) or very weak (one confirmation out of three) support.

In testing H2 and H4, respondents’ age and parents’ highest educational attainment and socioeconomic status were deemed as confounding variables, and gender was added. Career- and household-related variables can be treated as confounders because they precede work orientations in time and represent critical life events that influence work values. Therefore, we controlled for these variables while testing H2 and H4 in Model 4. Job-related variables were considered as intervening variables in Model 4. This specification may overestimate or underestimate the effects of work value orientations on income because a reciprocal causal relation exists between job-related variables and work value orientations. Therefore, we specified a new model (Model 5 in Table 3) that also controlled for job-related variables.

Models 1 to 5 assume homogeneous effects of the variables of interest. This does not necessarily need to be considered. In Section 2, we argue that the effects of gender on work value orientations may differ among groups. For instance, the difference between younger women and men may be smaller than that for older women because of a general trend in value convergence. With reference to income, heterogeneous effects can also be expected. For example, work value orientations may have different effects on men and women. Therefore, we analyzed Models 3 and 5 again, including interactions of gender with all the other variables (Models 3* and 5*).

General linear models (SPSS GENLIN IBM SPSS 2016) were used to estimate the parameters of the models. The significance level was set at p < 0.05. For convenience, the tables contain two-tailed error levels. For our one-tailed H1 to H4, we added the one-tailed error level, which was used for testing.

We report our findings in the following ways. Section 4.1 describes the results of Models 1 to 3 testing H1 and H3. Bivariate results were added for descriptive purposes. The results of testing the different operationalizations for extrinsic work value orientations and heterogeneous effects (e.g., Model 3*) are also reported. Section 4.2 portrays the results of testing H2 and H4. Descriptive results (trivariate results) and the results of testing the different operationalization for extrinsic work value orientation and heterogeneous effects (Model 5*) are included. Section 4.3 combines the results of Sections 4.1 and 4.2 and demonstrates the contribution of work values to explaining the gender wage gap. For this purpose, the independent variables were included step-by-step to demonstrate the contribution of the different kinds of independent variables. The first step included control variables in Model 1; the second included those in Model 2, etc.

3.4. Limitations

As discussed above, it was difficult to define the control variables for the analyses. We solved this problem by specifying different models and attempted to include different relevant control variables. However, we ignored some possible control variables. For example, it would be interesting to control for a partner’s income because extrinsic work values can become less important in the case of a partner with high income. Analyzing this expectation would imply concentrating on respondents in a partnership, which is beyond the scope of this study.

Apart from the problem of control variables, the cross-sectional data raises a causal problem because a reciprocal causal relation exists between work value orientations and income. However, this problem is less serious, and there are reasons to assume that value orientations influence income to a greater extent than vice versa (Vansteenkiste et al. 2007). Nonetheless, it would be preferable to have panel data to estimate the effects in both directions (income on work value orientations and work value orientations on income). Unfortunately, no panel data including socio-economic variables and value orientations are available for Austria.

In terms of reliability, the lower reliability of extrinsic work value orientations may lead to lower or insignificant effects on and of extrinsic values. Again, it might also be the case that they would display equally significant effects if they could be measured more reliably. Our second operationalization did not solve the problem of reliability but enabled an analysis of the consequences of a narrower understanding of extrinsic work values in the sense of career orientation. Another disadvantage is that we had no preference measurements for the value orientations. According to Vansteenkiste et al. (2007), an individual’s orientation toward one value is not relevant, but their preference for one value over other values is.

An additional limitation is that our analysis focused on a single country. Austria is a conservative-corporatist welfare state with a low degree of de-familialism and de-genderization. A strong gender segregation in the labor market and a neo-traditional division of labor between the sexes characterize Austria, where women are responsible for reproductive and care work and are employed part-time. Therefore, our results may hold for similar countries, such as Germany, but not for the other countries with differing characteristics, such as those in Scandinavia.

We have only analyzed one point in time, and therefore we were not able to track the gender wage gap’s progression over time like Böheim et al. (2021) or Fritsch, Verwiebe and Liedl (2019) in their analyses. Additionally, we did not differentiate between subgroups of employed persons in our analysis, so the specific situations of single-parents (Heitzmann and Pennerstorfer 2021; Riederer 2021) for instance, remains hidden.

4. Results

4.1. Gender differences in work value orientations

Table 4 shows gender differences in work value orientations. The data complies with H1. Women have significantly higher results than men for intrinsic work value orientations in the bivariate case and in the models.⁵ The results strongly support H1. The interaction analysis (Model 3*, results not shown) reveals one significant interaction. The gender differences in intrinsic work values were larger for persons without a partner than for those with a partner because men without a partnership evaluate intrinsic work values less than men in a partnership. The reason for this finding might be that men become more intrinsically orientated in a partnership as predicted by adaptive socialization theory (Lewis 2016; Oppenheimer 1988). Alternatively, intrinsically orientated men might in line with the general pattern of homophily (Lewis 1992; McPherson, Smith-Lovin and Cook 2001), particularly value homophily in our case, find a partner more easily because there is a better match with intrinsically oriented women.

Table 4. Intrinsic and extrinsic work value orientations by gender (estimated marginal means).

Analysis	Female	Male	Difference	p-value (Two-tailed)	p-value (One-tailed)
	Intrinsic work values
Bivariate	4.24	4.05	0.19	0.000	0.0000
Model 1	4.23	4.05	0.18	0.001	0.0005
Model 2	4.07	3.88	0.19	0.003	0.0015
Model 3	3.99	3.86	0.13	0.067	0.0335
	Extrinsic work values
Bivariate	4.00	4.10	−0.10	0.020	0.0100
Model 1	4.00	4.09	−0.09	0.021	0.0105
Model 2	3.88	3.90	−0.02	0.607	0.3035
Model 3	3.89	3.82	0.07	0.208	0.1040

Source: SSÖ (2016), only employed persons without a migration background, n = 905.

Remarks: Scale values for intrinsic and extrinsic work values range from 1 to 5, with 5 = very important and 1 = not important at all. Models 1–3 are described in Table 3.

By contrast, H3 applies only to the bivariate case and Model 1. After controlling for further variables (Models 2 and 3), the difference becomes insignificant. Hence, only very weak support exists for H3, which assumes gender differences in extrinsic work value orientations. The extrinsic value orientations were influenced by the other variables. The most important predictors in our analysis (Model 2, results not shown) were age, respondent's educational level, and place of residence. Employees under the age of 50 years, with a middle education (ISCED 3), and living in Vienna are more extrinsically orientated. Including other variables (Model 3, results not shown) reveals the effects of economic branch, occupation, and working time, where there are interaction effects with regard to the industry and working hours. For example, in female-dominated branches “Tourism” and “Health and Social Services” women are more extrinsically oriented than men. The analysis of interaction (Model 3*, results not shown) also points out that extrinsic working orientations are associated with working time only for women. In summary, the analysis of interactions brings several significant interaction effects. This result implies that gender differences in extrinsic work values depend on other factors. There were subgroups with large and others with no gender differences. In some subgroups, women have higher extrinsic value orientations than men and men report higher extrinsic work value orientations in other subgroups.

Using another operationalization of extrinsic work value orientations (see Section 3.2) leads to significant gender differences in extrinsic work value orientation in the bivariate case and Models 1 and 2, but gender is insignificant in Model 3. Hence, this alternative operationalization, which excludes security and concentrates on high income and advancement, provides weak support for H3.

4.2. Effects of work value orientations on income

Testing H2 and H4 for gross monthly income reveals a significant total main effect of intrinsic value orientations in the trivariate case and for all models (see Table 5). However, the effects point to a different direction than assumed in H2: persons’ income increases if they assess the value of intrinsic work values more positively. In Section 2, we consider that intrinsic work value orientations may have a positive effect on income via job satisfaction. Employees with intrinsic work value orientations are more satisfied. Therefore, they are more motivated, productive, and stay longer in a firm, which is honored with a higher income. Several studies have revealed that intrinsically orientated persons are more satisfied with their jobs (see Section 2.3). In our data, intrinsic work value orientation and job satisfaction were also positively correlated (r = 0.138, p(one-tailed) =0.000). In addition, job satisfaction was positively associated with income (r = 0.148, p(one-tailed) =0.001). However, the effect of intrinsic work value orientation on income cannot be explained by including job satisfaction as an intervening factor. The effect of intrinsic work value orientations remains statistically significant. This does not imply that the higher productivity of intrinsically orientated persons is not responsible for their higher income but underlines that further research is necessary on this topic to discover the causal link between intrinsic work value orientation and income.

Table 5. Effects of intrinsic and extrinsic work value orientations on monthly net income.

Analysis	Intrinsic work value orientation			Extrinsic work value orientation
	Effect	p-value (Two-tailed)	p-value (One-tailed)	Effect	p-value (Two-tailed)	p-value (One-tailed)
		Gross monthly income
Trivariate	225.40	0.000	0.0000	2.05	0.977	0.4885
Model 4	160.90	0.003	0.0015	94.17	0.153	0.0765
Model 5	128.23	0.008	0.0040	64.37	0.263	0.1315

Source: SSÖ (2016), only employed persons without a migration background, n = 619.

Remarks: Scale values for intrinsic and extrinsic work values range from 1 to 5, with 5 = very important and 1 = not important at all. Trivariate = intrinsic and extrinsic work values as independent variables, and income as the dependent variable. Models 4–5 are described in Table 3. Effects are estimated marginal means.

By contrast, extrinsic work value orientation has no significant effect on income in the trivariate case or in the two models. Hence, H4 must be rejected. The analysis of interactions (Model 5*) detected no significant interaction between gender and work value orientations on income. Therefore, we can assume that the above-described results hold for men and women; differentiating among subgroups of men and women is not necessary.

The alternative operationalization results in a significant effect of extrinsic work value orientations on income for Model 4. Model 5 and the trivariate analysis remain insignificant; therefore, H4 has weak support.

4.3. The relation of gender wage gap with work value orientations

Table 6 demonstrates how the different set of variables explain the gender wage gap in Austria. The gender difference without controlling for other variables has a value of EUR 824.6 for gross monthly income. Controlling for socialization-related variables of Model 1 (age and parent’s highest education level and socioeconomic status) did not reduce the gender wage gap, because there were no significant gender differences in these variables. A sharp decrease occurs after including the control variables of Model 2. The main reason for the decline is the duration of interruptions in occupation. There are large significant gender differences in this variable and the variable has a strong significant effect on income. In our sample, 29.4% of employed women report an interruption of four years or longer; among employed men, this percentage is 1.1%. The other variables did not contribute to reducing the gender wage gap either because they were not significant or not associated with gender. Another significant drop can be observed after adding the job-related variables to the control variables. This reduced the gender wage gap to a value of EUR 349.2. The main reasons for this are differences in working time, occupation, and economic branch. Technicians and associate professionals (ISCO major group 3), for example, have an average gross monthly income of EUR 2,347.5 and consist of 66.4% males. By contrast, for the service and sales workers (ISCO major group 5), the average gross monthly income is EUR 1,995.6 and the female share is 69.4%. However, men within each group also earn more. Similar patterns can be found for other economic branches. Finally, the average working time of women is 34.2 hours per week, whereas men report 42.9 hours a week.

Table 6. Gender wage gap explained by different set of variables.

Model	Absolute in EUR	p-value	in %	Absolute in EUR	p-value	in %
	Gross monthly income
Men	2,630.8	0.000	100.0%	2630.8	0.000	100.0%
Women	1,806.2	0.000	68.7%	1806.2	0.000	68.7%
	Wage gap in gross monthly income
Without control variables	−824.6	0.000	100.0%	−824.6	0.000	100.0%
After including control variables of Model 1	−811.9	0.000	98.5%	−811.9	0.000	98.5%
After including control variables of Model 2	−537.8	0.000	65.2%	−537.8	0.000	65.2%
After including control variables of Model 3	−349.2	0.000	42.3%	Included after work value orientations
Work value orientations	−379.7	0.000	46.1%	−594.6	0.000	72.1%
After including control variables of model 3	Included before work value orientations			−379.7	0.000	46.1%
	Contribution of work value orientations to wage gap
Effects of work value orientations	30.5	0.011	3.7%	56.8	0.003	6.9%

Source: SSÖ (2016), only employed persons without a migration background, n = 619.

Remarks: The control variables of Models 1 to 3 are defined in Table 3. Pw-value for two-tailed testing was examined because the hypotheses were undirected. Control variables may increase or decrease the gender wage gap. Effects are estimated marginal means.

Controlling for work value orientations increase the gender wage gap. This is mainly the case because women are more intrinsically orientated, and intrinsically orientated employees earn a higher income. The increase is small, but partly because the analysis excludes indirect effects of work value orientations on income via the job-related variables. If we change the ordering of the variables, the effect of work value orientations on gross monthly income increased from EUR 30.5 (3.7% of gender wage gap) to EUR 56.8 (6.9% of gender wage gap). These values change only slightly if we use the alternative operationalization of extrinsic work value orientations in the analysis.

5. Discussion

The aim of this study was to analyze whether work value orientations can explain gender wage differences in Austria. The analyses of gender wage differences have been mainly carried out by economists and concentrate on “objective” factors. To our knowledge, the role of work value orientations has not been systematically analyzed to explain the gender wage gap; this study aims to bridge such a gap. This study suggests that work value orientations can explain the gender wage gap if women are more intrinsically work-orientated (H1) and more willing to accept a lower income (H2), or/and if men are more extrinsically work-orientated (H3) and less likely to accept a lower income (H4).

Data from the Social Survey Austria 2016 were used to test these hypotheses. To consider the fact that different control variables were present, various models were applied in our data analysis. The analysis showed that employed women had more intrinsic work value orientations than employed men. Therefore, H1 was confirmed, but H3 was not. One explanation for the failure of H3 (men are more extrinsically orientated) might be the lower reliability of the measurement of extrinsic work values. We tried another operationalization, but we were unable to find a scale with higher reliability. Nevertheless, this alternative operationalization provides weak support for H3. This can be explained by the fact that the modified scale concentrates on career-oriented work values, such as high income and advancement, which are considered especially typical for men in gender stereotypes. Moreover, the results indicate that extrinsic work orientations are more strongly influenced than intrinsic work orientations by current social structures, especially by job-related factors, with gender becoming less important.

Regarding gross monthly earnings, intrinsic work value orientations have a significant influence on income. However, the effect of intrinsic work orientations goes in the opposite direction to that assumed by H2; therefore, H2 is proven to be false. Ceteris paribus, intrinsic work orientation leads to a higher income. H4 could not be accepted because no significant effect was present if our original operationalization was applied; the alternative operationalization also provided only weak support.

The results contradict the findings of Johnson (2001) and Johnson and Monserud (2012) for the US, where extrinsic work value orientations are associated with a higher income, and intrinsic work value orientations result in a lower income. In our opinion, the different social and cultural contexts are responsible for these divergences. Austria is a conservative-corporatist welfare state with a school system and labor market that are highly segregated by gender and the dominance of a neo-traditional division of labor between the genders. While value changes are observable, they have not yet altered institutional settings. These conditions explain why gender differences in intrinsic work value orientations are observable. In liberal countries, such as the US, a stronger need to adapt work value orientations to working conditions may exist, as well as a stronger importance of having a high income. In addition, fluctuation in jobs may be higher, so that an intrinsic work value orientation does not lead to a higher income via a longer stay in a job.

In summary, the fact that women have higher intrinsic work orientations lowers the wage gap. The wage gap in Austria would be larger if men and women had the same intrinsic work orientation. The gender difference in gross monthly income would – ceteris paribus – increase by about 3.7% to 6.9% if no gender difference existed in intrinsic work value orientation. Even though this amount is low, the findings are significant and substantially important.

Our study has several limitations. First, it is based on cross-sectional data; therefore, causal inferences are difficult. In addition, only absolute measures of work value orientations are available, and the reliability of the measurement of extrinsic work value orientation while being acceptable is low. Moreover, our study focused on a single country.

Nevertheless, our findings are of theoretical and methodological importance. Theoretically, our analyses demonstrated that the general value acquisition model is useful for empirical research. It allows us to deduce hypotheses and helps explaining differences within and between countries that are reported in the literature. Our findings demonstrate that work value orientations and their effects should be studied with respect to a cultural and structural context. However, more theoretical, methodological, and empirical work is necessary to explain the observed link between intrinsic work value orientations and income.

From a methodological viewpoint, our study underlines the importance of using panel data to solve causal problems. Moreover, the results suggest using a measuring instrument that can cover more dimensions of work value orientations with higher reliability. In line with other studies (see Section 2.1), our results indicate that differentiating between security and career aspirations within extrinsic work value orientations might be useful. Similarly, treating altruistic and social work value orientations as separate subdimensions of intrinsic work value orientation might be productive.

Finally, our results show that differences in the duration of interruption are an essential factor causing gender inequalities in wages. Wage differences between and within economic branches and occupations are further reasons associated with segregation in the educational system and the labor market. Therefore, gender policy must continue to focus on these factors to reduce the gender wage gap.

Overcoming these limitations and identifying the aforementioned theoretical and methodological aspects could be a fruitful task for further research. Work value orientations can enrich sociological, psychological, and economic analyses.

Acknowledgments

We would like to thank our anonymous reviewers for their valuable comments. The paper benefitted considerably from their suggestions.

Additional information

Funding

This study is part of a broader project on gender differences in reemployment and was supported by the funds of the Oesterreichische Nationalbank (Austrian Central Bank, Anniversary Fund, project number: 18231).

Notes on contributors

Johann Bacher

Johann Bacher is a full professor of sociology and head of the Department of Empirical Research (Institute of Sociology) at Johannes Kepler University Linz (JKU). His current research fields are research methods, youth unemployment and NEET youth, value orientations, and social inequalities. E-Mail: [email protected], JKU: https://www.jku.at/institut-fuer-soziologie/abteilungen/empirische-sozialforschung/team/johann-bacher; ORCID: https://orcid.org/0000-0002-8151-8922, Twitter: @BacherJohann

Martina Beham-Rabanser

Martina Beham-Rabanser is a senior researcher at the Department of Empirical Research (Institute of Sociology) at Johannes Kepler University Linz (JKU). Her current research fields are the sociology of the family, generations, and childhood, as well as the gender division of labor. E-Mail: martina.beham-[email protected], JKU: https://www.jku.at/institut-fuer-soziologie/abteilungen/empirische-sozialforschung/team/martina-beham-rabanser.

Matthias Forstner

Matthias Forstner is a PhD student at the Department of Empirical Research (Institute of Sociology) at Johannes Kepler University Linz (JKU). His current research fields include disability studies, research methods, and the digitalization of teaching. E-Mail: [email protected]; JKU: https://www.jku.at/institut-fuer-soziologie/abteilungen/empirische-sozialforschung/team/matthias-forstner.

Notes

1 Because of the differences in the way statistics are gathered in most countries, this article only describes the differences in income between two genders, male and female. Only recently, statistics and surveys have included information on trans- and intergender.

2 Similarly, Fritsch, Verwiebe and Liedl (2019) applied Blinder-Oaxaca decomposition for low-wage-employment risk and include human capital factors and labor market characteristics as explaining variables.

3 Providing a summary of the state of research on this topic is difficult because the studies used varying measurement instruments and examined different countries and subgroups. In addition, comparable and broad-based studies are missing because research in the 1990s concentrated on the effects of work value orientations or specific subpopulations, such as students, graduates, and professionals in certain fields or branches. For example, Frankel et al. 2006 studied Polish professionals; Johnson and Mortimer 2011 and Shevchuk, Strebkov and Davis 2018 concentrated on young people in the “new economy”; Marini et al. 1996 explored high school seniors; and Levey and Silver 2006 compared value orientations in Japan and the United States.

4 The following example illustrates this aspect: Variable X influences variable Y via the intervening variable Z. If variable Z is controlled in a statistical analysis, the effect of X on Y will become zero, and we would wrongly assume that X has no influence on Y.

5 Model 3 was significant because a one-tailed error level must be used.