Believing in equality: the effect of gender role ideology on labor market outcomes in the Netherlands

(1)

University of Amsterdam Faculty of Economics and Business

Bachelor Thesis

Supervised by Huaiping Yuan

“Believing in equality: the effect of gender role ideology on labor market outcomes in the Netherlands”

Geert van Bemmelen June 29th, 2020

(2)

Abstract. Inspired by recent increased attention for the gender wage gap, this paper explores

the relationship between gender role ideology and labor market decisions and their consecutive outcomes for working individuals in the Netherlands. Previous research has shown that a traditional gender role ideology can be associated with relatively more worked hours, but relatively lower income for men. This paper discusses the components of the gender wage gap, after which regressions are run using data from the LISS Panel to estimate the effect of gender role ideology on weekly hours worked and net monthly income for Dutch men and women. The results imply that Dutch men do not tend to put their ideology into practice, whereas there is a significant difference in hours worked for Dutch women, dependent on their ideology. There seems to be no significant relation between ideology and income for men and a small yet significant relation between the two for women. The inclusion of an interaction term to quantify the effect of parenthood again shows no significant effect for men, but a difference of 15 percent fewer hours worked by traditional mothers.

Introduction

Recent research by the Dutch Central Bureau of Statistics has shown that the gender wage gap has been decreasing in the Netherlands over the past decades (Van der Put et al., 2019). Contrastingly, a report by the World Economic Forum revealed the Netherlands dropping from the 27th spot in 2018 to the 38th spot in 2020 on the Global Gender Gap Index, being number 60 in the ranking regarding gender equality in economic participation and opportunity (World Economic Forum, 2018). These somewhat contradicting findings show that the wage gap has proven itself to be a persistent problem in the development towards gender equality in the Netherlands. Although it seems to be trivial to pay men and women the same salary for the same job, aforementioned data suggest that the problem is more complex than merely paying equal wages. Recent research using data provided by Uber on their US drivers, for example, has shown that a wage gap can even form when payoff is decided by a gender-blind algorithm, suggesting that men and women tend to behave differently on the labor market (Cook et al., 2018). The ones behaving most optimally are subsequently rewarded more than those who do not, explaining part of the wage gap. Aside from the potential problems related to the external validity of this research, its results could be a signal of a broader trend in society that explains, or at least partially so, the difference in earnings between men and women. It seems that traditional family values, where the husband is the breadwinner and the wife’s focus is on household tasks, have become less prevalent in the Netherlands in the past decades (Traag, 2020). If men and women put this shifting ideology into practice, one would expect that it

(3)

would also be visible in the data on the gender wage gap. However, the WEF shows that there are still substantial inequities in terms of labor market outcomes between Dutch men and women.

A number of studies in this field of research focus on the effect of parenthood on labor market outcomes of parents (see e.g., Bulanda, 2004; Glauber & Gozjolko, 2011; Kaufman & Uhlenberg, 2000). Furthermore, research on the effect of traditional gender role ideology has shown either a positive (Judge & Livingston, 2008) or a negative (Christie-Mizell, 2006) relation between traditional gender role ideology and earnings, depending on gender and/or race. Using data from the LISS (Longitudinal Internet Studies for the Social sciences) panel administered by CentERdata (Tilburg University, The Netherlands), this research paper will focus on the effect of traditional gender role ideologies on labor market outcomes, in particular on earnings and hours worked, of Dutch men and women. The first section will consist of an overview of relevant literature. Section two will elaborate on the sample data and method of analysis. The third section contains results and their implications. Finally, some concluding remarks, limitations and suggestions for further research.

Literature review

The Gender Wage Gap and Its Components

There can be several interpretations of the gender wage gap. In some cases, the gender wage gap is defined as the difference in average earnings between men and women, controlling for factors like education and hours worked. The other common definition of the gender wage gap is the difference in earnings between men and women that have the same occupation and would thus be expected to receive the same wage. This paper will focus on the former definition, which has slowly been declining over the past few decades in the Netherlands, according to data provided by the CBS, but has not yet been eradicated. Roughly speaking, the gender wage gap is often split into a part that is explainable and a part that is unexplainable by usual variables and thus believed to be solely due to gender-based discrimination (Goldin, 2014). In the coming paragraphs, I will elaborate on some of the explanatory variables.

1. Human capital

Firstly, there is the component that could be called skill, which is determined partly by genetics and partly by education. This component is part of the human capital approach, which generally speaking explains a large share of the difference in earnings between any two individuals, regardless of other personal characteristics. Educational attainment has often been

(4)

shown to directly influence the amount of earnings over a lifetime. Up until recently, there has been a gender difference in educational attainment tilted in favor of men in the Netherlands. But, according to a 2008 paper by van der Meer, from 2000 onwards women in the Netherlands have been on average more highly educated than men, meaning that they would be expected to have higher average earnings, which is not the case.

A second component that fits in the human capital approach, is that of a difference in labor force participation between men and women. This difference can be explained by preferences for the amount of time spent in the labor market. Women tend to spend less time on the labor market for several reasons, one of which, namely parenthood, will be discussed later. This means that women tend to gain less experience on the labor market and subsequently end up in lower paying jobs or sectors (Mihăilă, 2016). A final factor is one of selection, which is expressed in a difference in preference for the type of labor or occupation. In the Netherlands, this is noticeable in certain occupations within education and healthcare. Some functions within these social sectors, GP’s assistants or kindergarten teachers for example, have a majority of female employees and are relatively low-paid (CBS, 2019).

2. Discrimination

Another possible source for differences in earnings is discrimination. While this paper focuses on the differences between men and women, discrimination on the labor market can be happening to all kinds of groups of people, like majority and minority individuals, heterosexuals and homosexuals, religious and non-religious etc. These discriminatory practices can occur through several ways. Discriminative preferences held by (future) employers can play a role, as researched in a well-known paper by Goldin & Rouse from 2000. It showed the positive effect of introducing blind auditions for orchestras on the chance of success for female auditees, proving the existence of gender discrimination with conductors’ quotes that expressed a clear preference for hiring men (Goldin & Rouse, 2000). In other experimental research, it has been found that women are less likely to initiate negotiations about wage if their superior is a man and that women who do initiate negotiations are punished for it more severely by male evaluators than by female evaluators (Bowles et al., 2007). Several other studies have tried to estimate the effect of discrimination by conducting correspondence tests (see e.g., Acquisti & Fong, 2019; Bertrand & Mullainathan, 2004; Carlsson, 2011; Rooth, 2009; Tilcsik, 2011). A correspondence test entails sending out identical resumes with randomly assigned credentials that are irrelevant for a hiring decision but that do signal the fictional owner of the constructed resume to be part of a specific group. The results of those tests are often based on the callback

(5)

rates for individuals and, while callback rates do suggest that discrimination on the labor market exists, they are not synonymous for fully discriminatory practices. However, it is safe to assume that discrimination is a common issue on the labor market.

3. Child penalty

A component that is worth mentioning separately, even though it can be argued to be a part of preference for labor force participation is the so-called child penalty. Having a baby has been shown to negatively impact the earnings of women for the entire remainder of their professional life, and not so much the earnings of men. A study utilizing IVF treatments for Danish women as an instrument, estimated a 11-12 percent decrease in earnings for women after having children, whereas effects for their partners are significantly smaller (Lundborg et al., 2018). Part of this effect is due to reduction in hours worked when children are young, but in the long run lower wages cause the decline in earnings for mothers. Although seemingly unfair, this effect is not surprising. Biological differences mean that women experience more physical consequences of pregnancy and childbirth, meaning that it is logical that they reduce their hours worked temporarily. On the other hand, there are institutional tools that could even out this effect for mothers and their partners. In the Netherlands, parental leave schemes are tilted heavily in favor of the mother. Currently, according to a Dutch government website, a partner is entitled to one week of parental leave after birth, whereas mothers are entitled to a minimum of 10 weeks after birth and up to 20 weeks (Leave schemes, 2020). It is worth noting that there are plans to change this system somewhere in 2020. Introducing more equal paternal leave schemes could be a way to distribute the child penalty more equally over the mother and her partner, although that would mean that the partner would have to accept a loss of income for having a child (Selmi, 1999).

The child penalty is a consequence of two components of the gender wage gap discussed prior, namely that of labor force participation and sex-based discrimination. A pregnancy leads to a temporary absence on the job and the anticipation of a pregnancy could cause an employer to favor a male candidate when hiring. In the Netherlands, employers are therefore not allowed to ask a female candidate about whether she is planning to have children (Ministerie van Sociale Zaken en Werkgelegenheid, 2020).

4. Gender role ideology

Generally speaking, gender role ideology is defined as one’s opinions and beliefs about how individuals’ behavior regarding family and labor differs and should differ based on gender

(6)

(Harris & Firestone, 1998). The previously discussed components of the gender wage gap, but mainly the first two, play a role in the effects of gender role ideology on labor market outcomes. How individuals make decisions on their human capital accumulation, such as educational or occupational decisions, but also what individuals choose in terms of division of childcare or breadwinning can be influenced by opinions and beliefs on gender roles held by themselves or by others. The effect of others’ ideology can be substantial, as shown in the experimental research on wage negotiations, but estimating that is not in the scope of this paper. What is relevant, though, is the effect of ones (gender) beliefs on their behavior. Experimental research has shown how large this effect can be. When individuals in an experiment were asked to contribute to a discussion that is in their field of expertise, but where their gender does not align with the dominant gender in said field, they were less likely to contribute (Coffman, 2014). Furthermore, self-stereotyping can lead to a decrease in confidence which is observed to be larger for women than it is for men (Bordalo et al., 2019). These studies show that one’s beliefs can have an impact on behavior, meaning that they could impact labor market decisions and outcomes.

Regarding beliefs, the aim of this paper is to compare those on the more traditional end of the spectrum of gender role ideology with those on the non-traditional end. The effect of gender role ideology is especially relevant regarding parenthood, when it is common for either or both of the parents to reduce their hours worked in order to nurture their child(ren). According to a paper by Kaufman and Uhlenberg from 2000, the role of the father in these families can be on a spectrum ranging from “good-provider” to what they call “new fatherhood”. The former role describes the father as the one who is responsible for the economic provision of the household, whereas the latter dictates that a ‘good’ father shifts some of his worked hours from the professional setting to tasks in the household, including nurturing their children and household tasks. But childless partners will also have to make decisions on dividing household tasks, assuming that they are not fully outsourced. Traditionally, household tasks were mostly carried out by women, but non-traditional couples might have a more balanced division of these tasks. However, the division of household tasks is not a key variable in this research.

In the Netherlands specifically, traditional views on parental roles in the household have become less prevalent in the past decades, both institutionally and societally speaking (Traag, 2020). Some studies have looked at the effect of a traditional view on gender roles on labor market outcomes, although little to no research that is based on both Dutch and recent data is to be found. In 2006, however, Christie-Mizell studied race and gender differences of this effect in the United States and found significant negative effects of traditional gender norms on

(7)

earnings for all studied groups, except for white men. This implied that traditional views also negatively impacted the earnings of some men, in this case African American men, which was unexpected. Furthermore, Firestone et al. also find a negative effect of traditional gender role ideology on earnings for both men and women in a 1999 study. In contrast, the aforementioned study by Kaufman and Uhlenberg, also conducted with US data, showed a decrease in hours worked by non-traditional new fathers and an increase in hours worked for traditional new fathers, suggesting a positive effect on earnings for the latter group. Judge & Livingston also found a positive effect of traditional views on earnings in 2008. More recently, in 2011, Glauber and Gozjolko found that only highly traditional men worked significantly more than others after becoming a father. It has also been suggested that a working father is sometimes viewed by employers as more hard-working and more responsible, because they have to provide for their family (Kmec, 2011). This is again an indication for the effect that someone else’s beliefs can have on one’s earnings and hours worked.

The Netherlands as a Special Case

Together with aforementioned institutional and legislative aspects that aim to equalize the wage of men and women, there is also a specific element to the Dutch labor market that makes it an interesting case to study. In the Netherlands, it is relatively common to work part-time, and it has been so for a few decades. According to data provided by the European Commission, the Netherlands has steadily been the country with the highest percentage of part-time workers in Europe for a number of years with about 45 percent of part-timers, which is a good 10 percentage points above number two Switzerland (European Commission, 2020). A large part of this 45 percent is made up of women, of whom around 70 percent worked part-time in 2019 (CBS, 2020). With a labor market that is relatively attractive for working part-time, one would expect that individuals with non-traditional views would be more likely to align their stated preferences with their revealed preferences and reduce or increase their hours worked when needed, for example in the case of parenthood. On the other hand, the difference in average earnings between men and women means that men have on average higher opportunity costs and would thus be discouraged to reduce their hours worked.

Data and methodology

Sample

To study the effect of gender role ideology on earnings, this research will make use of data from the LISS (Longitudinal Internet Studies for the Social sciences) panel administered by

(8)

CentERdata (Tilburg University, The Netherlands). The LISS panel is a representative sample of Dutch individuals who participate in monthly Internet surveys. The panel is based on a true probability sample of households drawn from the population register. Households that could not otherwise participate are provided with a computer and Internet connection. A longitudinal survey is fielded in the panel every year, covering a large variety of domains including work, education, income, housing, time use, political views, values and personality. The specific data in this paper was gathered by the LISS panel in December of 2018 and in January and February of 2019. The original dataset consisted of 5,641 respondents. To make a meaningful comparison, some groups were excluded from the regression data. Firstly, 658 individuals that did not have a partner were excluded in order to only look at what influence gender ideology has on individuals’ behavior within a household. Secondly, individuals that did not have a relevant position within the household were excluded, for example children living at home with their parents or parents living with their adult children. Finally, individuals that did not have registered responses for the relevant questions on gender role ideology, earnings or hours worked were excluded. With these exclusions a data set of 1,495 observations remained, consisting of 742 males and 753 females. Table 1 in the appendix contains some further descriptive statistics on the groups of data.

Variables

Weekly hours worked. The LISS panel provides two measurements of respondents’ incomes. The first measurement is based on an inquiry on the worked hours that individuals are working contractually, whereas the second asks participants to estimate how many hours they actually work in an average week. The difference between these two variables is not significantly different from zero, and therefore the regressions will revolve around the contractual hours worked, as it is safe to assume that this is generally a reasonable representation of the hours worked by individuals.

Net monthly income. The income of the recipients is both measured and imputed in the LISS panel. The imputed variable is calculated through a process that is described thoroughly on the website of the LISS panel, but in short the process is meant to reasonably estimate the income of individuals for whom either gross or net income is missing or only indicated by a bracket of income, e.g. up to 500, 501 to 1000 etc. Individuals who had no information on income were dismissed from the data that was used in this paper. The distribution of the income variable is not significantly different from a normal distribution and the data does not show large outliers.

(9)

Therefore, the net income variable has not been transformed into a log variable. In the results section, however, the coefficients will also be explained in terms of percentage changes to help with understanding the effect sizes.

Gender role ideology. The measurement of gender role ideology is based on fourteen questions from the ‘Values and politics’ section of the LISS panel. Table 2 shows all the questions related to gender roles that were answered by the participants. Questions 1 to 7 and 12 to 14 were to be answered on a 5-point scale, ranging from fully disagree to fully agree. Questions 8 to 11 asked participants to specify what kind of job they think a mother should be able to have in several situations regarding the age of their children. In this case, filling out a 1 means an individual thinks a mother should be able to have a full-time job, 2 a part-time job and 3 means they think a mother should be able to have no job at all. In order to make all responses to the questions have the same implication, the scores for some questions were reversed. This has been done for the questions in table 1 that have an asterisk at the end.

To make a distinction between a traditional and a non-traditional view on gender roles, I turn to the methodology used by Glauber and Gozjolko in their 2011 paper. The response values for all fourteen questions were added up to create a total score for each participant, ranging from 14 to 54 and where a higher score means a more traditional stance on gender roles. The average score was 27.77, with a standard deviation of 6.66. In this paper, an individual is considered to have a traditional view when their total score is higher than one standard deviation above the mean, which is a score of 35 or higher. This methodological choice is based on a threshold effect that is found for women in the data. Although the effect is not as prevalent for men, the threshold has been applied for them too for the sake of making a comparison. When measured like so, 15.72 percent of the sample, or 235 individuals, is considered to have a traditional view on gender roles, of which 153 are male and 82 are female. Since it is very plausible that traditionality is not as binary as this approach makes it seem, a robustness check will be conducted to make sure that the dichotomous approach does not distort the results too much. This robustness check will be explained in more detail in the section on methodology.

Controls. The LISS panel data also provides variables containing background information on the participants. Age is one of the variables that is relevant in this regression, because of the correlation between age and earnings. The same reasoning holds for educational attainment, as has been mentioned in a previous section. Data on educational attainment is measured in six

(10)

educational categories by the LISS Panel, rather than in years of schooling. The lowest category in the data contains individuals with less than a high school degree. The increasing categories contain individuals who have finished intermediate secondary educations (VMBO, in Dutch), individuals with a degree in higher secondary education (HAVO/VWO), individuals with an intermediate vocational education degree (MBO), individuals who have finished higher vocational education (HBO) and finally individuals who have finished a university degree.

Furthermore, it seems sensible to control for the number of children the respondent has that still live at home, since it is more likely that individuals with children living at home will adjust their hours worked due to need for childcare. The age of the youngest child is also used as a control in the regression, because older children will generally speaking need less time in terms of care. For this variable a few categories were made, based on age and subsequently on what kind of school they attend. The first category contains children younger than 4 which do not attend school. The second category is from 4 to 12, when they attend primary school. The third category contains children in the age of 12 to 16 when they attend high school. The fourth category, containing children older than 16, is omitted because of collinearity. Finally, weekly hours worked and net monthly income were used as control variables when they were not being used as dependent variables because of the high correlation between the two. This is logical, because more hours worked will usually lead to relatively higher income and a higher income (due to a higher hourly wage) could be an incentive to increase the hours worked.

Methodology

The two labor market outcomes of interest for this paper are net monthly income and weekly hours worked. The regressions in this paper are ordinary least square regressions and will thus using either of the aforementioned measurements as the dependent variable. Following the approach taken by Firestone et al. in their 1999 paper, regressions were conducted separately for men and women in order to look at how a traditional gender role ideology correlate with labor market outcomes for the two groups separately.

First of all, table 3 in the appendix shows the predictors of a traditional gender role ideology. The results of this table have also been calculated by a simple linear model, with the traditional view dummy as the dependent variable and age, gender, education, number of children living at home, the age of the youngest child, income and dummies for first- or second-generation immigration backgrounds as predictors.

To continue, the first set of regressions have weekly hours worked as the dependent variable. The purpose of this is to find out the average difference of weekly hours worked between

(11)

individuals, depending on their view on gender role ideology. Based on all previously discussed research, the hypothesis would be that men with a traditional view would work more hours than their non-traditional counterparts, as they would fit in the good provider model, and that women with a traditional view, on the contrary, would work fewer hours since they would spend more hours running the household and taking care of the children. In the second set of regressions in model 1, the correlation between a traditional gender role ideology and earnings is analyzed. It is important to note that, contrary to those that had no information on hours worked, individuals that reported their worked hours to be equal to zero were included in this regression. This was the case for 27 individuals. According to what has earlier been called new fatherhood, the income distribution between men and women could be expected to be more equal for non-traditional individuals than for the more traditional. In the analysis, that would imply a positive correlation between the variable indicating traditionality and earnings for men and a negative correlation between the two variables for women. Needless to say, all previously mentioned controls were included in both models.

Furthermore, model 2 is included to test the robustness of the measurement for gender role ideology in the first model. Because of the difference in threshold effects between men and women, this model uses a different approach to measure traditionality. Model 2 includes the same four regressions as those in model 1 but with a different independent variable. Every individual’s score on the 14 gender role ideology questions were standardized to having a mean of 0 and a standard deviation of 1. The sum of those 14 values is then used as a measurement for traditionality, instead of the binary variable in model 1. This means that the coefficients in these regressions can be interpreted as the effect on the dependent variable when the total score for traditionality is increased by one standard deviation. If the results of model 1 and model 2 are not significantly different, it is safe to assume that the choice for the cutoff in the binary approach is not the main cause of the results.

Model 3 will add an interaction term between traditional gender role ideology and being a parent to the base regressions in model 1. The aim of this term is to isolate the difference between traditional individuals with children and traditional individuals without children. According to the theoretical discussion in earlier sections, it can be expected that traditional individuals with children would have a larger incentive to actually put their ideology into practice, meaning that this coefficient would indicate a larger decrease in hours worked for women than for men and perhaps even an increase in hours worked for men.

(12)

Findings

Results

Table 3 contains results on the predictors for having a traditional gender role ideology. There is a small yet significant negative relation with age and a traditional view, with the likelihood of having a traditional view declining by 0.8 percent for every year respondents gain in age. Furthermore, gender is a substantial predictor for having a traditional view. Women are approximately 17 percent less likely to have a traditional view on gender roles than men. Additionally, educational category is significantly related to traditional views. For every step up in educational category, respondents are 2.3 percent less likely to have a traditional view on gender roles. No significant relation between number of children living at home and a marginally significant relation between the age of the youngest child with a traditional view has been found in this dataset. The relation between net monthly income and traditionality is significant, but almost negligible due to its size, of around 0.003 percent, implying an average 3 percent decrease in likelihood of having a traditional view for every 1000 euros of additional net monthly income. Finally, it seems that having a first-generation immigration background does have a significant relation with traditional views on gender role ideology. Respondents that fall into this category are about 14 percent more likely to have a traditional view than individuals with a ‘Dutch’ background.

Turning to the main regression models, table 4 in the appendix shows the results of the first model. The first section of model 1, referencing to the left side of table 4, shows that men with a traditional view on gender role ideology work on average more weekly hours than their non-traditional counterparts, although the effect of 1.66 hours is small when comparing it to the average hours worked for males in this sample of 35.63, making it a change of 4.6%. Additionally, the effect is only marginally significant, with a p-value of 0.024. For women, however, the effect is larger and more significant. Traditional women work 2.31 hours less per week than comparable women with a non-traditional view. Comparing it to the 25.65 hours worked by female respondents on average, the change equals 9%. Moreover, the effect is quite significant with a p-value of 0.008. To put things in perspective, the difference in average hours worked between non-traditional men and women is approximately 9 hours per week, whereas the difference for more traditional men and women is approximately 26 hours. Another interesting outcome is the effect of number of children living at home on the hours worked of both men and women. Although again marginally significant, in general the number of children living at home has a negative correlation with hours worked for female respondents and a positive effect for male respondents.

(13)

The results on the right side of table 4 show that a traditional view on gender role ideology has no significant impact on men’s wages. For women there seems to be a slightly significant negative relation between traditionality and earnings. Furthermore, a significant positive correlation between age and earnings exists for both men and women, which is in line with expectations based on models of lifetime income distribution. The significant and positive relation between education and income is also in line with what theory would predict. The number of children living at home and the age category of the youngest child shows no significant relation with earnings for both gender groups. The positive correlation between weekly hours worked and net monthly income is significant for both male and female respondents, however it is worth noting that the coefficient is almost twice as large for women as it is for men in this sample.

To see if the results hold up to a different measurement of traditional gender role ideology, table 5 uses the sum of all 14 standardized survey answers to create a total test score, which is then used as the independent variable. The standardized variable logically has a mean of approximately zero, with a standard deviation of 8. The results of the regressions with weekly hours worked as the dependent variable are very similar in significance to the results in table 4. The difference in correlation between the two tables is negligible for almost all control variables. The coefficient for the standardized test score is now insignificant for men. On the other hand, the coefficient for women has become significant at the 0.1 percent level. Its effect size has become smaller, but less so when comparing it with the binary cutoff method in table 4. To compare, in this case a standard deviation of 8 above the mean would mean a decrease in hours worked of 8 times 0.186 hours, which is 1.5 hours. The same consistency holds for the regressions that revolve around net monthly income. Most coefficients do not change significantly, but significance for the coefficient of the standardized test score increases for women in combination with a slightly smaller effect size. Although not perfect, it seems that the choice for a binary variable for gender role ideology does not drastically influence results. Finally, it is interesting to see whether individuals with children are more likely to change their behavior on the labor market according to their ideology. The results of the regressions exploring this can be found in table 6. An interaction term between traditional ideology and a dummy variable for being a parent or not is included in the regressions. Again, for men the results are not significant when looking at the correlation between the interaction term and weekly hours worked. For women, however, the interaction term is somewhat significant and signaling towards a decrease of almost 4 hours per week on average. The coefficient for the interaction term regarding net monthly income is insignificant for both men and women.

(14)

Implications

Firstly, the correlation between gender and traditional gender role ideology is in line with previous research, which also found that women are more likely to have a more liberal approach to women’s roles (Eagly et al., 2004). Secondly, the relation between immigration backgrounds and traditional gender role ideology is substantial. In a study by Alesina et al. that was published in 2013, it was found that specific historical aspects of cultures can have an effect on gender inequality and gender norms in the present. Although specific information on respondents’ background, such as country of origin of (grand)parents was not included in the analysis of this paper, the results can still be interpreted as showing that certain cultures are more likely to have traditional views on gender roles.

Overall, the results from the regressions are most often significant and substantial for women. Although traditional men tend to work a bit more on average than non-traditional men, the significance of the results is lost when an interaction term for being a parent is added. In this case, the lack of significance can be a result in and of itself. It can be interpreted as the fact that traditional and non-traditional men do not tend to work different numbers of hours. Men that are on the non-traditional side of the ideology scale thus do not tend to put their ideology into practice by reducing their hours worked to take care of household or child-related tasks. Women, on the other hand, do tend to work less when they are categorized as having a traditional view. Especially mothers with a traditional view show this effect with an average difference of approximately 15 percent of the hours worked, when compared to similar women in the dataset.

Finally, the effect of education on earnings is smaller for women than it is for men, yet it is highly significant for both. This could be due to gender-based discrimination on the labor market, as discussed previously. The higher degree of difficulty faced in hiring procedures or a superior’s smaller willingness to cooperate in wage bargaining could be a reason for this difference.

Conclusions

Mainly, this paper intended to estimate the effect of traditional gender role ideology in the context of the gender wage gap, by looking at its effects on the labor market decisions and outcomes of Dutch men and women. In conclusion, it seems that a traditional view on gender roles has some of the expected effect. That is, traditional women do tend to work fewer hours

(15)

than non-traditional women, but non-traditional men do not seem to work significantly less than traditional men. Furthermore, a traditional view has not been found to have a significant effect on the earnings of men, but it does seem to cause slightly lower earnings for women. Other predictors, such as gender and cultural background, are in line with previous research on gender role ideology and labor market outcomes.

It is important to note some limitations to this research as well, together with some suggestions for further research. The exclusions of several groups of individuals due to the theoretical foundations of this research caused a relatively small sample size to remain. Although it did not prevent some results from being significant and although the percentage of individuals with a traditional view was in line with larger sample groups in other research, it would be interesting to see how the effects change when a larger sample of Dutch respondents is studied. Reverse causality could also be an issue of the approach taken in this paper. This paper has tried to argue on a theoretical basis that the correlations found in the results are likely to be the sign of a certain causality. However, it could be the case that the number of hours worked or income influence one’s gender ideology instead of the other way around. For example, it could be the case that being a mother that reduces hours worked because of nurturing causes one’s gender role ideology to become more traditional, due to getting into contact with other (traditional) stay-at-home mothers more often. Taking an instrumental variable approach would get rid of this issue but finding a suitable instrument would not be easy. Furthermore, it could be interesting to isolate the effect of new parenthood by making use of time-series data and examining the change of labor market decisions after a first child is born. This shift from a between to a within design could estimate the effect of traditional views on labor market outcomes more precisely. Finally, elaborating on the effect of one’s gender role ideology and how they express their ideology in everyday behavior could be an interesting angle to approach the topic from. More qualitatively, studying behavior in an experimental setting and focusing on different occupational categories and the gender distributions within these categories, for example, would be a way to explore and possibly quantify the effects of stereotyping on performance of men and women in certain occupations.

(16)

Bibliography

Acquisti, A., & Fong, C. (2019). An experiment in hiring discrimination via online social networks. Management Science.

Alesina, A., Giuliano, P., & Nunn, N. (2013). On the origins of gender roles: Women and the plough. The Quarterly Journal of Economics, 128(2), 469-530.

Bertrand, M., & Mullainathan, S. (2004). Are Emily and Greg more employable than Lakisha and Jamal? A field experiment on labor market discrimination. American economic review, 94(4), 991-1013.

Bordalo, P., Coffman, K., Gennaioli, N., & Shleifer, A. (2019). Beliefs about gender. American Economic Review, 109(3), 739-73.

Bowles, H. R., Babcock, L., & Lai, L. (2007). Social incentives for gender differences in the propensity to initiate negotiations: Sometimes it does hurt to ask. Organizational Behavior and human decision Processes, 103(1), 84-103.

Bulanda, R. E. (2004). Paternal involvement with children: The influence of gender ideologies. Journal of Marriage and Family, 66(1), 40-45.

Carlsson, M. (2011). Does hiring discrimination cause gender segregation in the Swedish labor market?. Feminist Economics, 17(3), 71-102.

CBS. (2019). In één derde van beroepen op hoogste niveau is meerderheid vrouw. Retrieved 20 May 2020, from https://www.cbs.nl/nl-nl/nieuws/2019/46/in-een-derde-van-beroepen-op-hoogste-niveau-is-meerderheid-vrouw

CBS. (2020). Meer dan de helft werkt voltijds. Retrieved 22 April 2020, from https://www.cbs.nl/nl-nl/nieuws/2020/08/meer-dan-de-helft-werkt-voltijds.

Christie-Mizell, C. A. (2006). The effects of traditional family and gender ideology on earnings: Race and gender differences. Journal of Family and Economic Issues, 27(1), 48-71.

Coffman, K. B. (2014). Evidence on self-stereotyping and the contribution of ideas. The Quarterly Journal of Economics, 129(4), 1625-1660.

Cook, C., Diamond, R., Hall, J., List, J. A., & Oyer, P. (2018). The gender earnings gap in the gig economy: Evidence from over a million rideshare drivers (No. w24732). National Bureau of Economic Research.

Eagly, A. H., Diekman, A. B., Johannesen-Schmidt, M. C., & Koenig, A. M. (2004). Gender gaps in sociopolitical attitudes: A social psychological analysis. Journal of personality and social psychology, 87(6), 796.

(17)

European Commission. (2020). Part-time employment rate. [Data file]. Retrieved from: https://ec.europa.eu/eurostat/databrowser/view/tesem100/default/bar?lang=en

Firestone, J. M., Harris, R. J., & Lambert, L. C. (1999). Gender role ideology and the gender based differences in earnings. Journal of Family and Economic Issues, 20(2), 191-215. Glauber, R., & Gozjolko, K. L. (2011). Do traditional fathers always work more? Gender

ideology, race, and parenthood. Journal of marriage and family, 73(5), 1133-1148.

Goldin, C. (2014). A grand gender convergence: Its last chapter. American Economic Review, 104(4), 1091-1119.

Goldin, C., & Rouse, C. (2000). Orchestrating impartiality: The impact of" blind" auditions on female musicians. American economic review, 90(4), 715-741.

Harris, R. J., & Firestone, J. M. (1998). Changes in predictors of gender role ideologies among women: A multivariate analysis. Sex Roles, 38(3-4), 239-252.

Judge, T. A., & Livingston, B. A. (2008). Is the gap more than gender? A longitudinal analysis of gender, gender role orientation, and earnings. Journal of applied psychology, 93(5), 994. Kaufman, G., & Uhlenberg, P. (2000). The influence of parenthood on the work effort of

married men and women. Social forces, 78(3), 931-947.

Kmec, J. A. (2011). Are motherhood penalties and fatherhood bonuses warranted? Comparing pro-work behaviors and conditions of mothers, fathers, and non-parents. Social Science Research, 40(2), 444-459.

Leave schemes. (2020). Retrieved 2 April 2020, from https://business.gov.nl/regulation/leave-schemes/

Mihăilă, R. (2016). Female labor force participation and gender wage discrimination. Journal of Research in Gender Studies, 6(1), 262-268.

Ministerie van Sociale Zaken en Werkgelegenheid. (2020). Welke vragen mogen niet gesteld worden tijdens een sollicitatiegesprek?. Available at: https://www.rijksoverheid.nl/onderwerpen/gelijke-behandeling-op-het-werk/vraag-en-antwoord/welke-vragen-mogen-niet-gesteld-worden-tijdens-een-sollicitatiegesprek [Accessed 02 April 2020]

Lundborg, P., Plug, E., & Rasmussen, A. W. (2018). Can Women Have Children and a Career?. Rooth, D. O. (2009). Obesity, attractiveness, and differential treatment in hiring: a field

experiment. Journal of human resources, 44(3), 710-735.

Selmi, M. (1999). Family leave and the gender wage gap. NCL Rev., 78, 707.

Tilcsik, A. (2011). Pride and prejudice: Employment discrimination against openly gay men in the United States. American Journal of Sociology, 117(2), 586-626.

(18)

Traag, T. (2020). Opleiding en werk: twee generaties vrouwen vergeleken. CBS. Available at:

https://www.cbs.nl/nl-nl/achtergrond/2020/10/opleiding-en-werk-twee-generaties-vrouwen-vergeleken [Accessed 02 April 2020]

Van der Meer, P. H. (2008). Is the gender wage gap declining in the Netherlands?. Applied Economics, 40(2), 149-160.

Van der Put, A., Chkalova, K. & Van Gaalen, R. (2019). Jonge moeders dragen steeds meer bij aan gezinsinkomen. CBS. [online] pp. Available at: https://www.cbs.nl/nl-nl/achtergrond/2019/04/jonge-moeders-dragen-steeds-meer-bij-aan-gezinsinkomen

[Accessed 29 March 2020]

World Economic Forum, 2020. Global Gender Gap Report 2020. [online] pp.8-14. Available at: http://www3.weforum.org/docs/WEF_GGGR_2020.pdf [Accessed 31 March 2020]

(19)

Traditional Nontraditional Male

(N = 176) (N = 90) Female (N = 566) Male (N = 663) Female

Variables M SD M SD M SD M SD Hours worked 34.05 _16.59 17.79 _12.65 36.30 _13.96 26.54 12.49 Net income 2342.19 _762.62 1192.18 _709.90 2569.21 _957.89 1628.12 775.63 Part-time 0.11 0.32 0.61 0.49 0.17 0.38 0.70 0.47 Controls Age 49.89 _10.24 44.86 _10.41 49.57 _12.07 46.37 12.23 Parent 0.61 _0.49 0.62 _0.49 0.51 _0.50 0.56 0.50 Number of children 1.14 1.10 1.28 1.15 1.00 1.16 1.02 1.09

Age of youngest child at home 17.91 10.40 15.51 10.25 18.72 11.13 18.60 11.18

Education

Less than high school 0.03 _0.18 0.01 _0.11 0.02 _0.14 0.01 0.12

Intermediate secondary education 0.20 _0.40 0.20 _0.40 0.11 _0.31 0.13 0.34

Higher secondary education 0.05 _0.21 0.09 _0.28 0.09 _0.28 0.09 0.29

Intermediate vocational education 0.37 _0.49 0.44 _0.50 0.25 _0.43 0.26 0.44

Higher vocational education 0.25 _0.43 0.22 _0.42 0.35 _0.48 0.33 0.47

University 0.11 _0.31 0.04 _0.19 0.20 _0.40 0.16 0.37

(20)

Table 2. Questions used to generate indication of traditional or nontraditional gender role ideology.

Question Mean answer

1. A working mother’s relationship with her children can be just as close and warm as that of a non-working mother. * 1.82 2. A child that is not yet attending school is likely to suffer the consequences if his or her mother has a job. 2.26

3. Overall, family life suffers the consequences if the mother has a full-time job. 2.34

4. Both father and mother should contribute to the family income. * 2.37

5. The father should earn money, while the mother takes care of the household and the family. 1.72

6. Fathers ought to do more in terms of household work than they do at present. * 2.72

7. Fathers ought to do more in terms of childcare than they do at present. * 2.70

Do you think that women, under the circumstances described below, should be able to have a full-time job, a part-time job, or no job at all?

8. If she has a baby (a child younger than 1 year). 1.94

9. If she has a child that does not yet attend school. 1.80

10. After the youngest child starts primary school. 1.51

11. After the youngest child starts secondary school. 1.26

12. A woman is more suited to rearing young children than a man. 2.53

13. It is actually less important for a girl than for a boy to get a good education. 1.42

(21)

Traditional ideology Age -0.0079** (0.0024) Female -0.1766*** (0.0266) Education -0.0234* (0.0097) Number of children 0.0147 (0.0121) Age of youngest child living at home 0.0052* (0.0026)

Net monthly income -0.0000**

(0.0000) First generation immigration background 0.1393***

(0.0417) Second generation immigration background 0.0233

(0.0473)

Constant 0.8805***

(0.1104)

Observations 1118

Adjusted R-squared 0.059

Standard errors in parentheses * p<0.05, ** p<0.01, *** p<0.001

(22)

Table 4. Regression coefficients from model 1, including regressions on weekly hours worked (left) and net monthly income (right).

Weekly hours worked Net monthly income

Men Women Men Women

Traditional ideology 1.655* -2.310** -77.52 -133.5* (0.730) (0.866) (77.59) (64.54) Age -0.109*** -0.147*** 11.99*** 5.407** (0.0292) (0.0251) (3.089) (1.898) Education -0.680** -0.0782 249.6*** 189.6*** (0.240) (0.249) (23.90) (17.18) Number of children 0.839** -0.698* 26.39 -38.67 (0.297) (0.287) (31.68) (21.36) < 4 years 0.160 -1.430 202.2 62.30 (1.125) (1.022) (119.0) (76.10) 4-12 years -0.496 -2.297** 86.17 101.5 (0.999) (0.888) (105.9) (66.21) 12-16 years -0.516 -0.816 118.1 108.9 (1.162) (1.085) (123.1) (80.64)

Net monthly income _{0.00212*** 0.00798***} (0.000340) (0.000397)

Weekly hours worked 23.76*** 44.15***

(3.815) (2.197) Constant 37.50*** 22.09*** -49.21 -576.7*** (1.982) (1.805) (256.2) (144.0) Adjusted R-squared 0.085 0.492 0.170 0.538 Observations 742 751 742 751

(23)

Table 5. Regression coefficients from model 2, including regressions on weekly hours worked (left) and net monthly income (right) with standardized scores as robustness check.

Men Women Men Women

Standardized score 0.0610 -0.186*** -5.021 -11.29*** (0.0369) (0.0373) (3.904) (2.796) Age -0.110*** -0.150*** 12.10*** 4.946** (0.0292) (0.0248) (3.090) (1.887) Education -0.674** -0.182 246.5*** 180.1*** (0.243) (0.247) (24.18) (17.25) Number of children 0.841** -0.618* 27.37 -34.02 (0.298) (0.284) (31.69) (21.22) < 4 years 0.112 -1.095 205.9 79.68 (1.127) (1.013) (119.0) (75.61) 4-12 years -0.522 -2.130* 92.97 106.4 (1.004) (0.875) (106.1) (65.48) 12-16 years -0.464 -0.628 119.5 117.5 (1.164) (1.073) (123.0) (80.02)

(24)

Table 6. Regression coefficients from model 3, including regressions on weekly hours worked (left) and net monthly income (right) with an added interaction term.

Men Women Men Women

Traditional ideology 1.129 0.0831 -100.3 -201.8* (1.130) (1.371) (119.8) (102.0) Traditional x Parent 0.876 -3.905* 38.09 112.4 (1.437) (1.738) (152.3) (130.0) Age -0.107*** -0.151*** 12.06*** 5.533** (0.0293) (0.0251) (3.105) (1.904) Education -0.676** -0.0501 249.8*** 188.6*** (0.241) (0.249) (23.92) (17.22) Number of children 0.778* -0.518 23.76 -43.67* (0.314) (0.297) (33.40) (22.13) < 4 years 0.198 -1.409 203.9 61.92 (1.127) (1.020) (119.3) (76.11) 4-12 years -0.501 -2.386** 85.91 104.4 (1.000) (0.886) (106.0) (66.31) 12-16 years -0.550 -0.796 116.6 108.4 (1.164) (1.082) (123.4) (80.66)