Gender equality: Education, Employment and Segregation Factors
in the European Union
J.M.E. Vennemann1 Supervisor: T.M. Stelder
University of Groningen 28-11-2012
Abstract
This paper investigates three gender gaps that may have a relation with economic growth: education, labor force participation and segregation indices for field of study and occupations. A panel study is performed with five models for the European Union. Two short-term models are conducted with annual data for the period 1970-2010 and for the period 1993-2010. Two long-term models are based 5-year averages data for the period 1970-2010. Both models allow variation over time and one of them measures the delayed impact of education and labor force participation on economic growth. The final short-term model is conducted for the period 1998/1999-2010, to measure the impact of segregation indices. The results imply that only tertiary education has a positive impact on economic growth. Even a higher gender gap in tertiary education could generate more growth in short-term. A result from increased participation of females here. For labor force participation, only significant results are found related to the overall level in the short-term. According to segregation indices, especially robust estimates are found regarding segregation indices for occupations.
JEL-codes: I25, J16, 015
Keywords: economic growth, gender inequality, education, labor force participation, segregation,
European Union
1
2
1. Introduction
Kofi Annan, Secretary General of the United Nations from 1997-2006, once stated: “Gender equality is more than a goal in itself. It is a precondition for meeting the challenge of reducing poverty, promoting sustainable development and building good governance” (United Nations, 2005). According to Annan, reaching gender equality is a prerequisite for achieving two of the eight Millennium Development Goals devised by the United Nations. The first one of these is Millennium Development Goal 2, achieving universal primary education: ‘ensure that, by 2015, children everywhere, boys and girls alike, will be able to complete a full course of primary schooling’. The second is Millennium Development Goal 3, promoting gender equality and empowering women: ‘eliminate gender disparity in primary and secondary education, preferably by 2005, and in all levels of education no later than 2015’. These goals address gender equality in rights and accessibility to education and labor force, which is not only to promote equity, but they are necessary for economic development too (World Bank 2005).
Gender inequality is generally defined as the biological and social differences between males and females. According to Blau, Feber and Winkler (2010) gender inequality is multidimensional and has an impact in social, cultural and economic areas: the first two deal with human rights and the final aspect is concerned with social status. Obviously; one social aspect is the prevailing traditional division of roles, where men have a job and women take care of the household. A lot of economic consequences can be investigated too. In this research the relationship between gender inequality, education and employment and economic growth will be explored and discussed. Depending on how the term ‘gender inequality’2 is defined, the answer to this relationship will vary and therefore assumptions will be made.
The focus is on three aspects of gender inequality. Referring to figure 1 (next page), the first inequality is between males and females in getting access to education. The participation of men and women in education can be estimated at the primary, secondary and tertiary level. According to the human capital theory proposed by Schultz (1961) and further developed by Becker (1994), the main objective of education or training is to increase the level of knowledge and skills of the workforce, which would consequently raise the marginal productivity of the workforce. Therefore, human capital, including education, has been emphasized as a critical determinant for economic progress. Basic literacy and numeracy will contribute to enhancing the productivity of lowly skilled people, while analytical skills will augment the productivity of highly skilled people. Thus if the productivity of
3 workers will increase and the output of goods and services will be stimulated. Within this context, equal opportunities between men and women merit particular attention. Equal access to education is also partly an essential condition for equal opportunities on the labor market.
Regarding employment we have two different aspects. First, the unequal participation rate in the labor force. As mentioned above the traditional division of roles is an important explanation for the differences between men and women in participation. Even though the common belief is that women should be concerned more about housekeeping than having a job, history shows that several factors have probably had a positive impact on the participation rate of women. First of all, as women obtained more education, the wage rate increased and they were more likely to work outside their home. Secondly, changes in values and attitudes towards the traditional model contributed. Like Smith (1974), modernization theorists focus on how institutions (including schools) transform values, beliefs and attitudes. A crucial contribution to this process is that if more people participate in education, a greater level of modern individuals will be attained and the perception of women in the workforce changes. Summing up by looking at figure 1, the participation of females could also have an impact on human capital and therefore on economic growth.
The final inequality lies in the hours of work and the payment that men and women receive. Women still dominate in taking care of the family responsibilities at home. That is why a balance between work and private life is harder to find for women than for men. Lack of childcare facilities and the costs that this entails force women to exit the labor market. The labor force participation for
Figure 1. Gender gaps and the impact on economic growth
Male Female
Gender gaps
Primary, Secondary and Tertiary Education 1. Inequality in participation in
education
Labor force participation 2. Inequality in labor force
participation
Labor force working hours and
remuneration 3. Inequality in working hours
and remuneration
4 women with dependent children is 62,4 percent, while this is 91,4 percent for men with dependent children (European Commission, 2012). A solution for both taking care of family responsibilities and be able to perform a job lies in part-time work. One-third of women works part-time, while only 8 percent of men does in Europe (European Commission, 2012). As a consequence, their contribution to human capital will be lower. The European Commission defines the gender pay gap as the relative difference in the average gross hourly earnings of women and men. Of course, earnings have an impact on their participation rate. If the earnings are high, it is more likely they will look for a job if they don’t have one. Earnings will not be discussed in further detail. However, the earnings will be taken into account indirectly since they influence the participation rate, which has an impact on human capital.
Increased global economic integration has caused several reforms of markets and government regulations. According to the European Commission (2007), the six founding countries of the European Economic Community (EEC) adopted an equal treatment of payment legislation as early as 1957. As the EEC expanded, so did the legislation on equal treatment. According to World Atlas Gender Equality (2012), gender inequality still persists and predominantly to the disadvantage of women (European Commission, 2012). Reducing inequality is important since it can improve the well-being of women, and simultaneously of the whole economy.
Several studies have investigated the relationship between gender inequality in education and economic growth. Gender inequality in secondary education in particular seems to have significantly negative effects for economic growth. Moreover, the results suggest that gender inequality in employment have a negative impact on economic growth. These studies differ in design, but the most important thing to notice is that the research areas are quite different.
Before presenting some earlier studies, it is worthwhile mentioning some theoretical terminology. First, gender parity is explained as the aim of allowing males and females equal participation in education and the workforce. Gender equality is broader than gender parity. Gender equality is to gain access and participation in education and the labor force, as well as having an equal treatment and having the same benefits on social, economic and political aspects3. Gender parity, therefore, is a first step toward gender equality. Gender equality enhances equal rights for men and women. An European Institute for Gender Equality is established in 2006 and became fully independent in 2010. The European Commission Strategy is to promote equal treatment of both sexes and to combat sex discrimination in Europe.
3
5 Secondly, it is generally known that the economic growth of a country depends on a wide range of factors. The gross domestic product (GDP) per capita refers to the final products and services within a country for a specific period. Countries with a low GDP level per capita often have a higher growth rate due to the catch-up effect4. This means that in the long term, countries will converge to the same level of GDP per capita, in other words, countries that start off poor will grow faster.
It is plausible that there is a correlation between gender inequality and economic growth. A higher proportion of women in the labor force will boost production, the starting up of new businesses, thereby stimulating innovations. Therefore it is likely that less inequality between the sexes will have a positive impact on GDP levels per capita and will result in higher economic growth rates too. At the same time, it can be assumed that the correlation holds in the opposite direction as well. In a rich country, more gender equality can be expected to exist, and it is likely that more regulations are implemented to reduce inequality. If a higher productivity is achieved women will be needed too, and educational opportunities should be guaranteed. Finally, less inequality will boost economic growth, but it could be that the reverse correlation holds too.
Summing up, it is important to establish that gender gaps in education, employment and payment have different effects, but are correlated too. On the one hand, gender gaps in education could lead to inequalities in employment, since employers prefer educated people. On the other hand, if there is a barrier for women to participate in the labor force or if the payment for of women is low, parents might decide that the gains of education of girls are far less than those of boys. So education of females would not be stimulated, and gender gaps in education are a consequence.
As mentioned before the evidence of this research could be important in view of the two Millennium Development Goals regarding gender equality. Abu-Ghaida and Klasen (2004) estimated that the costs of failing to achieve these goals by 2015 would be a 2,4 percent increase in malnourished children and a 1,5 percent higher mortality rate in the 0-5 age group. Although the access to primary education is more of a problem in developing countries than in the EU, it seemed to be interesting to investigate this nonetheless.
The current research topic has been extensively studied in the past for large parts of the world, in particular for the developing countries. However, the unequal treatment of men and women could have an impact on the developed countries too. This topic has not yet been the subject of empirical research in the European Union. Since the EU is an important guide for European
6 countries to establish policies, it would be interesting to know what the impact of gender inequality is in this region. Presumably the countries will use different tools but they will certainly have some in common. By looking at the Eurozone into more detail, this research could help provide solutions how to improve enrolment into education5. Of course the evidence could have implications for the empowerment of women in the labor force too. Results could indicate whether women get enough opportunities to be able to participate in education and the labor force. It will also be interesting to know if women indeed have an impact on the economic growth of a country. Besides this, tertiary education is new in relation to gender inequalities too, since it has not been investigated before. Simultaneously, the study performed will be a panel study here. This is different from previous studies, since they conducted cross-country regressions. Furthermore, a new potential relation on gender inequality will be proposed. The first one suggests that there is a relation between the distribution of men and women in fields of study and economic growth. The second one investigates the relationship between distribution of both sexes regarding occupations and economic growth. The reasoning behind this relationships will be explained in more detail in section 2.
The remainder of the paper is organized as follows. Section 2 discusses the relevant theories and literature on gender inequality and economic growth. It outlines the hypotheses. Section 3 elaborates the methodology in order to create a growth regression. Section 4 specifies the variable selection and their measurements. Section 5 gives descriptive statistics for the participation of men and women in education and the labor force. Section 6 presents and analyzes the empirical results. Section 7, finally, gives a conclusion and identifies areas for future research.
2. Relevant theories and literature
The introduction briefly explained the three known gender inequalities that may be related to economic growth. Several studies have offered important findings on this relationship and this section will discuss it in more detail. The main topics are the inequality in education and the inequality in labor force participation. Related to these main topics, an additional issue is taken into account, namely the impact of an inequality in segregation of men and women in different fields of study and occupation.
2.1 Education
Economic theorists have explored extensively the relation between education and economic performance. The first model is the neo-classical or exogenous growth models, in which the growth rate is a result of determinants from outside the model (known as Solow-Swan model). This model
5
7 will converge to a steady state growth model, which depends on technological progress and the rate of labor growth. Another model of growth was featured by Lucas, Romer and Weil (1992)6, by using a broader measure of capital, namely physical and human capital. This results in a model which is more in line with reality than the Solow-Swan model. Endogenous growth theorists argue that investment in human capital, research and development and knowledge contributes to greater technological progress, which, in turn, stimulates economic growth. Much of the literature on the general rate of return of human capital found a positive relationship of increased education on economic growth too (Barro and Lee (1994), Knowles and Owen (1995), Easterly and Levine (1997) and Sachs and Warner (1995)). A confirmation of this relationship is provided by Stevens and Waele (2003). Education is needed to create new technologies and to use the available technology appropriately, which will eventually stimulate technological progress. According to these studies, education is an important factor in increasing human capital to stimulate economic growth. This results in the following hypothesis.
H1. A higher overall level of (primary, secondary and tertiary) education has a positive effect on economic growth.
Empirical data therefore show that there is a positive correlation between education and economic growth. However, opinions on whether the effects of education are the same for males and females differ. For example, Barro and Lee (1994) and Sala-i-Martin (1995) find a negative coefficient on female education in relation to economic growth. However, a lot of other researchers have disputed these findings on the grounds of empirical problems and multicollinaerity (Knowles et all 2002), which is the problem of a high correlation between two variables. Moreover, studies with cross-sectional data reveal two things for gender inequality in education. First, gender inequality in education is higher in low-income level countries7. Secondly, gender inequality tends to be high in countries with low total literacy rates, for example argued by Dollar and Gatti (1999).
Klasen (2002) shows that there are direct effects of a gender bias in education on growth, because lower participation of females in education lowers the average level of human capital and eventually has a negative direct effect on economic growth. It is important to note that economic growth is measured in terms of GDP. Such a measure excludes the value of women’s household activities at home. The exclusion of this unpaid work underestimates the value of women’s total contribution, according to Klasen (2002).
6
Heijdra, B.J. (2009), ‘Foundations of modern macroeconomics’, second edition, Oxford University Press, Oxford.
8 Figure 2 shows that gender inequality in employment results in a selection-distortion factor at the labor force level, as there is less labor productivity, compared to cases where there is no gender gap in education. So due to the lower labor productivity, the amount of human capital decreases and consequently economic growth drops. Additionally, a less productive workforce results in a lower rate of return to capital, which reduces the investment rate. Lower labor productivity therefore also indirectly reduces economic growth via investment.
The participation of females in education has positive indirect effects as well. A direct externality effect is called the environment effect (see figure 2 again). Female participation has an impact on the quality and quantity of education for the next generation, because a well-educated mother can provide support and a better general environment for her children. Given the levels of education for men, lower female education has negative consequences for the overall quality of education of the population. Klasen (1999) argues that equal access to education at the household level has positive effects on the quality of education. For example, siblings will stimulate and support each other in reaching educational success. Similarly, couples will strengthen for life-long learning. If a men is learning and working his whole life, he will stimulate his women to do the same. In conclusion, more educated women in the household improves the intellectual environment at home, which leads to more labor productivity and so economic growth.
Barro and Lee (1994) and Klasen (1999 and 2002) arrived at opposite conclusions on the impact of female secondary education. Both conducted a cross-section analysis and did not take into account the endogeneity of the gender variable. In other words, they did look at the causation that female education could boost economic growth, but this causation could hold the other way around as well. Moreover, it is important to study the determinants and calculations of gender inequality.
Figure 2. Selection and distortion and environment effects
9 A cross-national study carried out by Dollar and Gatti (1999) support the findings of Klasen (1999) by using cross-national data for over 100 countries, five-year growth intervals and two-stage least squares estimation. In contrast to earlier studies, economic growth is measured by GNP per capita8. They find that a higher participation of females in secondary education has a positive relationship to growth, while more participation of males in secondary education lead to smaller growth rates. Dollar and Gatti (1999) and Klasen (1999) therefore argue that the prejudice of underinvestment in education of females results in a misallocation of education resources and thus lowers growth. Secondly, Dollar and Gatti (1999) find a correlation between a GNP per capita and a gender bias, which is strongly convex. The improvement in the participation rate for female of secondary education decreases as countries go from low to middle incomes, while from middle to high income levels, an increase in income has a strengthening effect in female education. Dollar and Gatti (1999) argue that “one plausible explanation of this relationship is that there are market failures that hinder investment in girls and that these failures diminish as countries develop”.
Hill and King (1995) also examined the relationship and regressed levels of aggregate output on gender gap, female secondary school enrollment and several control variables by a cross-national research. Their study showed that women’s educational levels have a significant positive impact on GNP and that a higher gender gap in education is associated with lower levels of GNP. This means that for given levels of female education, higher levels of male education reduce GNP.
Knowles et al (2002) investigated with a cross-national study the impact of educational gender gaps on GDP per capita, using a Solow model by including the male and female educational levels as separate variables of production. The most important result is that female education has a positive impact on labor productivity. If there are declining marginal returns to education, average human capital will be lower if girls have restrictions to education and boys will have higher levels of education. Economic performance can be boosted if boys and girls have equal opportunities. By taking into account the defects of causation and collinearity of panel data, they conclude that gender gaps in education are impediments for economic progress.
A contribution to the literature on gender inequality in education and economic growth in developing countries is proposed by Baliamoune-Lutz and McGillivrary (2007). The focus was on two different variables, namely the ratio of females to males in primary and secondary education and the ratio of literate 15-24-year-old females to males. Using panel data for African and Arab countries, it shows that inequalities in literacy have a negative effect on economic growth. Higher gender
8
10 inequalities have an even stronger effect in Arab countries. In addition, more open economies have more growth thanks to trade, but this is coupled with greater inequalities. Greater inequalities in literacy therefore have a positive effect here. Again there is statistical evidence that the enrolment of females in secondary education has positive effects on economic growth. Women’s increased level of education should also cause sex differentials in the rate of return on education to decrease.
Before presenting the hypotheses on gender inequality in education, a note must be made on pre-primary and primary education. First on pre-primary education, the purpose of which is to create an opportunity for children to play and interact with peers with some form of professional guidance. In many developing countries, early childhood is becoming a major priority. A body of research has shown that students who have attended pre-primary education perform better in primary education and are more likely to move on to higher education levels. At the same time, pre-primary education allows parents to work part-time or full-time, which may have an impact on the participation rate of labor. As pre-primary education is not sufficiently developed in all EU countries and the impact on economic growth is not sure jet, it could give wrong implications if it were taken into account here. In contrary, primary education is compulsory for every child (UNESCO 2012) for developed countries, like the EU. In all probability this makes the gap for primary education between both sexes very small in these countries and therefore this gap is not relevant in this research here. To confirm this the descriptive statistics of primary education will be shown.
However, according to the literature described above, secondary education has a lot of common estimates to be observed. The role of secondary education is important in economic growth, especially secondary education for both males and females seems to have a significant impact according to previous studies. This leads to the following hypothesis.
H 2. A lower gender gap in secondary education has a positive effect on economic growth. A contribution to the existing literature is made in this research by looking not only at primary and secondary education levels, but by taking into account tertiary education too, which is a relatively unexplored area of research. This results in the following hypothesis.
H 3. A lower gender gap in tertiary education has a positive effect on economic growth.
2.2 Labor force participation
11 the growth rate is determined by population growth, capital accumulation, productivity and technological progress. Barro and Sala-i-Martin (1995) confirm that conditional convergence holds in this model, which is that economies converge towards a steady state if they are almost similar9 countries. This convergence is a result of the technological progress and the labor force growth. An increase in the labor force will lower the capital per worker and the output per worker will be lower. So with too much labor, an increase in the labor force could reduce economic growth. On the other hand, a small initial amount of labor could increase the productivity and the total output.
Based on this model it is hard to determine the direct effect of the labor force participation on economic growth. Another limitation to this model is that the process of how technological progress is reached is not explained. This pitfall has created endogenous growth theories, which take technological progress or knowledge accumulation as a determinant from within the model. One of these endogenous models is of Uzawa (1965) and Lucas (1988), which assumes that technological knowledge augments labor. In other words, knowledge could have a positive effect on labor productivity. In this model there are two stocks that can be accumulated, namely the stock of physical capital and the stock of knowledge. It is assumed that with a high level of knowledge it is difficult to generate additional stock of knowledge. One trade-off which is faced is between the optimal level of labor force participation and the educational sector. Raising the number of people in the educational sector will increase the stock of knowledge, but it decreases the number of people available for production. According to the different exogenous and endogenous theories, the direct impact of the labor force participation rate could be controversial. To determine the relationship between labor force participation and economic growth the following hypothesis will be tested.
H 4. A higher labor force participation rate has a positive effect on economic growth. Figure 3 (on the next page) illustrates that higher female education has a strong social effect on child mortality and fertility. As mentioned above, well-educated mothers can generate a better environment for their children. Holtz et al (1997) argue that economic growth leads to higher return to human capital, which leads to a reduction in fertility rates, smaller families and investment for each child increases. If women get more education, it is argued that the population growth will reduce due to lower fertility rates, since the age of marriage is raised, and the time between pregnancies increases.
9
12 Bloom and Williamson (1997) investigated that through several influences on the so-called demographic transition effect, lower fertility rates have an impact on economic growth. First of all, a lower fertility rate reduces the youth dependency burden. This is the ratio of the number of youths (people of ages of 14 years and younger) to the number of people in the working-age population (people between 15 and 64 years). At the same time, the total dependency burden is reduced, which is defined as the ratio of the number of youths and the elderly (persons of ages 65 and older) to the number of people in the working-age population. This implies that aggregate savings in the economy will increase and thereby economic growth is stimulated. Secondly, for a limited time the share of the labor force becomes larger, due to the earlier high population growth. This larger productive workforce will boost investment demand. The higher domestic savings and the increase in the investment level will therefore both boost economic growth. A third argument is related to fact that education increases women’s participation in the labor force. Specifically, it is asserted that educated females will have a higher earning prospect and therefore they have an incentive to seek a job. Simultaneously, women will have more occupational aspirations. Of course, it also changes the traditional division of roles between men and women, and it makes women more eager to work.
The results generated by Gümbel (2004) are in line with Dollar and Gatti (1999), the variable of gender biases in employment is the most influential variable here. In the analysis some expectations were made about the variables and the impact on economic growth in the future. Contrary to what was expected, secondary education was not the most influential factor in predicting growth. This variable might be important in generating a high GNP, but it does not have any influence on achieving growth in the future. It is important to notice the content of the sample size, which consists of industrialized and democratic countries. So promoting more gender equality in secondary education would have a small impact on higher growth for these countries. However, as mentioned
Figure 3. Demographic transition effect
13 above, promoting more female participation in the labor force would create more overall labor force and thereby generate more growth. In line with this literature, Klasen and Lamanna (2003) provide estimates that a gender gap in education lowers growth in the Arab countries. At the same time, they provide information that the gender gap in employment is even higher and has more impact.
Nonetheless, with links on gender biases and economic growth it is important to take into account the inequality in activity rates for both sexes. The EU has set a growth strategy for the coming decade, which is to reinforce and reach high levels of employment, productivity and social cohesion. According to the European Council (European Commission 2011), 75% of the 20-64 year-olds should be employed. Over the last few years the rate of female labor participation has increased consistently. In line with this the following hypothesis will be tested.
H 5. A lower gender gap in labor force participation has a positive effect on economic growth.
2.3 Gender wage gap
The inequality in working hours and remuneration is often used in research about gender inequality. It will be explained here, but will be omitted in the final research, because the main topics are gender inequality in education and labor force participation.
Using a cross-national study, Cavalcanti and Tavares (2007) examined an economic growth regression with endogenous variables of savings, fertility and labor force participation. This paper is in line with those variables, but the current paper does not examine the differences in payment due to gender. Their results show that an increase of 50 percent more gender inequality in payments for paid work leads to a direct increase of 65 percent in hours worked at home by females. This eventually leads to a 25 percent lower output per capita compared to a situation of no wage gap. On the contrary: if women earn a higher payment, fertility rate reduces, since women have less time to take care of the family responsibilities. However if women get a higher remuneration, an incentive to work is created. Therefore women will work more hours which will produce more output per capita.
14 exporting products which are manufactured with an intensive use of female labor. It was concluded that countries that took most advantage of the low wages for women were the fastest growing ones.
2.4 Gender biases and other factors
To give a total view of research about gender inequality and to get an idea of democracy as a control variables, these other studies will be discussed below.
An interesting research worth mentioning was conducted by Plantega et al (2009). Different indices of gender inequality were developed, although an index especially for the EU was missing. Based on Fraser (1997), they composed an index of four dimensions: equal sharing of paid work, money, decision-making power and time. The index is constructed in such a way that the value indicates the actual distance from a situation of full equality. The empirical results show that full equality is still a long way off. Finland, Sweden and Denmark display the highest overall performance. The southern countries Greece, Cyprus, Malta, Spain and Italy perform rather poorly. However, aggregating different variables into one index could result in measurement errors, since the variables can differ in their impact on economic growth, and it is impossible to determine these impacts separately with one index.
Mitra, Bang and Biswas (2012) provide a working paper with two gender biases. In contrast to several earlier studies, it includes five distinct variables regarding gender equality. The first dimension is a gender bias regarding access to economic opportunities, including variables such as access to schooling, literacy and fertility. The second gender bias is related to the variables participation in the labor force and political participation. Their results show that only access to economic opportunities is significant and an increase in access will boost growth. This results support the earlier findings on public education and economic growth. In addition to the two dimensions, they included four variables about institutional characteristics to investigate the interplay between gender equality and institutions. The results of Mitra et al (2012) illustrate that the level of democracy, transparency of the government and credibility of a regime have a significant and positive impact on growth, while security of civil society is not significant.
15 Klasen and Wink (2003) argue that the level of democracy plays a critical role in creating women’s access to education. With more democracy, the bargaining power of women within the household increases. This not only strengthens the position of women, but it also creates better opportunities for the next generation, as women will stimulate their children, and especially their daughters, to be properly educated. However, owing to the fact that women will thus be enrolled in education longer, marriage will be postponed, which might result in lower fertility levels.
According to Acemoglu, Johnson and Robinson (2005), including institutions in a growth regression is important, because institutions shape the incentives of key economic actors. Institutions have an influence on the investment in physical and human capital. Geographical and cultural factors have an impact on differences between countries in economic growth, but it is the institutional differences which are the major sources. Institutions divide the economic outcomes among physical and human resources, in other words they are responsible for the income division among individuals. Therefore, the structure of political institutions impacts on the gender outcomes for the access to education and the labor force.
2.5 Segregation indices
In line with the different kinds of gender inequality mentioned above, in this research two new inequalities related to education and labor force will be investigated. These two new relations are contributions to the existing literature. The first one suggests that there is a relation between the distribution of men and women in fields of study and economic growth. The reasoning behind this relation is that even though the participation of women in education has increased tremendously, there are fields of education where the participation of women, compared with men, is already steadily higher.
Gertler and Alderman (1989) point out an argument which is indirectly in line with this diversification. They mention that a lower investment in education for girls than for boys could be a result from a lower return from schooling girls. Of course this can only be the case when labor of males and females are imperfect substitutes. And probably this is the case, because males have different innate abilities than females. They therefore develop an interest in different fields of study, namely those that are the most suitable according to their innate abilities. A counter-argument to this could be that men and women are complementary to each other. Since both sexes have different innate abilities, this creates different insights into different fields of study. This argument is in favor of a more equal distribution among men and women in fields of study. Both sides may be correct and this will be tested by the following hypothesis:
16 Of course, field of study has an impact on gender segregation by occupation. In a situation of gender equality, work will be rationally distributed between the sexes. Consequently, the most suitable person will perform the job, regardless of the gender of a person. It could be stated that there are jobs in which the participation of men is higher than that of women, for example in manufacturing and engineering jobs, which possess qualities or characteristics considered typical of or appropriate to a man. As a result of these typical male attributes, the productivity in these industries will be higher when more men are employed. The same holds true for women, who will probably have a higher participation in the teaching and service industries. However, diversification can lead to more productivity too. This argument has important implications for increasing the total productivity of labor. So segregation in occupations is valuable if it leads to an efficient productivity of labor. The European Commission (2009) reported that a problem can arise with strong segregation, since labor shortages for sex-typed occupations are hard to solve within the country10. With a high demand for labor for this specific occupation a bottleneck could result and cause inefficiencies in the labor market. According to Miller et all (2004) de-segregation, less sex-related occupations and a more equal distribution could be a solution to solve the problem of labor shortages. This results in the following hypothesis.
H 7. A more equal distribution in occupations has a positive effect on economic growth.
3. Methodology
There is considerable theoretical support for investigating the impact of gender gaps in different dimensions on a country’s economic growth. In this section, data will be presented for growth, education, labor force participation and segregation for different countries.
The relationship of gender inequality and economic performance depends on the characteristics of countries; Dollar and Gatti (1999) mentioned that civil rights and freedoms, religion or regional preferences all have an impact on gender inequality. Obviously, some criteria must be applied in the selection of countries in the sample, but it is difficult to control for all the influences explained by Dollar and Gatti (1999), as the subject of the current paper is education and employment inequality. Seguino (2000) stated that it is important to choose a sample that consists of countries in a similar stage of development and with similar economies. Consequently, the first criteria is selecting countries that are industrialized. The second criteria is that they should be part of the EU, since the findings will be used and discussed to answer the question whether there are policies to improve the situation of inequality in the Eurozone. A third criteria regarding Klasen and
10
17 Wink (2003) is the level of democracy, since it plays a critical role in creating access to education for females. It is a fact that all members of the EU are democracies11. Since data is not available for all variables, several data sets are used. Tables 1 and 2 provide information about the countries of the EU and their abbreviations.
[Insert tables 1 and 2]
By using these data sets, which contain different countries, it is possible to regress several different cross-country analyses in five different models:
1. Short-term model one is based on annual data for the period 1970-2010. 2. Short-term model two is based on the annual data for the period 1993-201012.
3. Long-term model three is based on average data across five-year periods for the period
1970-2010. A lot of economic research uses a five-year average, life for example Durlauf et al
(2005). Average data across five-year periods allows for variation over time and gives no misleading information about the long-term growth. Since the time that it takes for changes in the explanatory variables to have impact on the dependent variable is not specified here. 4. Long-term model four is based lagged values of average data across five-year periods for
the period 1970-2010. To determine the delayed impact of the education and labor force
variables on economic growth.
5. Short-term model five is based on data for the period 1998/1999-2010. This final model measures the impact of the segregation, data is only available for the years after 1998 for occupations and after 1999 for education fields.
The cross-country studies have shown that lower levels of inequality will result in higher growth. Therefore, governments that create policies and regulations to reduce inequality could improve economic growth. While cross-country regressions do support these findings, they do not directly take into account the possible endogeneity of the explanatory variables (as fertility rates, investment etcetera). To solve this problem and to evaluate the relationship between gender bias and economic growth of countries across time, a panel estimation is used. Other potential problems could be serial correlation, since yearly rates incorporate short-run fluctuations, which could create
11 According to the report Freedom in the World (2010): all European Union countries are covered in the
category of ‘Free’. Based on the average of civil and political rights they are qualified as ‘electoral democracies’ satisfying four criteria. One criteria is a fully competitive political system with multiple parties. The second is adults suffrage, through which citizens can express their political preferences. The third is contested elections on a regular base. The fourth is significant information about political parties is provided to the public by media and campaigning.
12
18 serial correlation from business cycles. Using the first differences of the variables this problem could be solved.13 Another potential problem could be heteroskedasticity, when the standard errors are non-constant over time. A solution for these standard errors is using White’s diagonal coefficient variance method. Following the literature according to the standard regression in panel data, the economic growth regression is as follows:
𝑦, = 𝛽∆𝑋, + 𝜃∆𝑍, + 𝑛 + 𝑢 + 𝜀 (1)
The dependent variable 𝑦, is the growth rate of real (inflation-adjusted) per capita gross
national product (GDP). The column vector 𝑋, includes all explanatory variables that determine
income, some of them are endogenous. Explanatory education variables are primary, secondary, tertiary education and labor force participation for hypotheses H 1 and H 4. Explanatory inequality variables are gender inequality in Gross Enrolment Rates in primary, secondary and tertiary education, gender inequality in labor force participation and segregation indices. These inequality variables are for all hypotheses H 2, H 3, H 5, H 6, H 7. The row vector 𝑍, includes all control
variables, which are initial GDP, investment rate, trade openness, fertility rates, population growth and level of democracy. All variables will be explained in further detail below. The partial regression coefficients 𝛼, 𝛽 and 𝜃 determine the extent to which a variation in the variables has an impact on the GDP of a country. The variable 𝑛 the country-specific fixed effects controls for country variant time invariant effects which may have impact on dependent variable economic growth. The variable 𝑢 the time-specific effects controls for country invariant time specific effects.14 The variable 𝑛 are the country dummies and the final variable 𝜀 is the error term.
4. Measurements and specifications
In this section the variable selection will be explained in further detail. Table 3 gives an overview of the variables, the abbreviations, the descriptions and the sources.
[Insert table 3]
Variable selection 4.1 Dependent variable Economic performance
A major debate is currently ongoing on the best possible benchmark for economic performance. Mankiw (2006) states that most research uses Gross Domestic Product (GDP) as the best indicator to
13 Wooldrige (2002) procedure is used in each regression to test for serial correlation in the residuals. The
procedure begins by estimating parameters for each regression. Giving these results the residuals are kept. The regression will be performed again including the lagged values of the residuals. Serial correlation is determined by a positive significant coefficient. Results show no serial correlation in any of the regressions.
19 provide information about the economic wealth of a country. In this study GDP is used to assess the strength of a country’s local economy and to compare it with other countries. As mentioned earlier, GDP represents the total production of goods and services by all producers in a country plus taxes and minus subsidies not included in the value of the products (World Bank 2012). GDP per capita is a better measure than total GDP, as it also takes into account the population size. Without this correction the picture would be incorrect. Other studies used GNP per capita, which measure a GDP plus the incomes earned by foreign nationals domestically. However, data of GNP have been discussed based on several grounds, such as quality, accuracy and comparability of the variable. The biggest problem of this variable is that it is hard to determine the incomes that are earned by nationals abroad, since they are differently recorded on official accounts. For example, if non-marketable goods and activities are taken into account by one country but omitted in another country, the comparability is weakened. Since the data set is based on industrialized countries, this problem is not that relevant.
4.2 Overall level of human capital 4.2.1 Education
Accumulation of human capital is an important part of economic progress; this includes capabilities and skills that are embodied in the ability to perform labor. Investment in human capital can be accounted differently, but generally it is connected to development in education. Additionally, several measures of human capital related to education can be used. In general, stocks of education are represented by completion, in other words, average years of schooling. In this research, however, the access opportunities or the investment in human capital are important, therefore the Gross Enrollment Rate (GER) in the different levels of education are used15.
The purpose of this rate is to show the general level of participation at a given level of education. The GER is ‘the number of actual students enrolled in a level of education, regardless of age, divided by the official school-age population corresponding to the given level of education in a given school year’. A high GER means a high degree of participation, whether students belong to the official age group or not. The achievement of GER to almost 100% indicates that a country has enough capacity to provide education for all of its school-age population.
As a consequence, the GER can exceed 100%, which makes sense since children that are enrolled can be over-aged (or under-aged), due to late (or early) school entrance. School repeaters of a level of education contribute to a higher GER too. To solve the problem of the enrolment of over- or under-aged children the Net Enrolment Rate (NER) can be used. This rate measures only those
20
students that belong to the relevant predefined age group for the particular level of education, therefore the NER can never exceed 100%. In other words, the NER is calculated by dividing the number of students of a particular age group, for example primary education, by the number of children in the population of that age group. The available data for the NER, however, is extremely low compared to the GER and since the indicators are almost substitutable, the GER is used here. In line with this, the different GER levels of males and females are used separately too.
4.2.2 Labor force participation
The overall level of the labor force participation rate is investigated in relation to economic growth too, because labor is needed to generate products and services. The labor force participation rate is the proportion of working-age people (those between 15 and 64 years old) from the total population. Those people are economically active, which means that everybody either is employed or is looking for employment.
4.3 Measures of gender inequality 4.3.1 Education
To determine the gender inequalities16 in education, the absolute differences in Gross Enrollment Rates (GER) are calculated. This gives the following gap variables17:
𝐺𝐴𝑃𝑆𝐸𝐶 = |𝐺𝐸𝑅 − 𝐺𝐸𝑅 | × 100%
𝐺𝐴𝑃𝑇𝐸𝑅 = |𝐺𝐸𝑅 − 𝐺𝐸𝑅 | × 100%
4.3.2 Labor force participation
For labor force inequality the absolute differences between activity rates are used18. This gives the following equation for this inequality:
𝐺𝐴𝑃𝐿𝐹 = |𝐴𝑐𝑡𝑖𝑣𝑖𝑡𝑦 𝑟𝑎𝑡𝑒 − 𝐴𝑐𝑡𝑖𝑣𝑖𝑡𝑦 𝑟𝑎𝑡𝑒 | ∗ 100%
Note for both inequalities the absolute relative differences will be calculated too. Since the absolute differences could differ widely between countries, therefore a correction for total enrolment rate makes it more valuable to compare inequalities between countries.
16 Since primary education is compulsory for every child the gap between sexes is nihil, which is confirmed in
the descriptive statistics later on. Therefore this variable is not taken into account here.
17
For the other three variable secondary and tertiary education there are some lacking GERs for countries in several years. However for every country there are preceding and following years, so the average of these are taken for the lacking years.
18
Activity rate are calculated as:
21
4.3.3 Indices of dissimilarity
Another variable which is relatively new in relation to studies on gender inequality and growth is the index of dissimilarity (ID)19. This indicator is used to describe the segregation among several fields of study or occupations. In other words, it provides information about the distribution of men and women and could give incentives to change the level of segregation to get an equal distribution. The ID index varies between 0 and 100 in percent, and a change in field of study or occupation is related to males or females. So a change in the level of female employment has an impact on both the structure of the occupation and a change in the composition of men and women within the occupation. Finally, it results in an overall change of segregation (European Commission 2009).
Segregation education
In relation to education the ID is measured to evaluate the distribution of the various education fields. Evaluating this ID for fields of education has probably not been done before, which makes it a contribution to the existing literature. The education fields that are included are general programs (1), education (2) humanities and arts (3), social sciences, business and law (4), science (5), engineering, manufacturing and construction (6), agriculture (7), health and welfare (8), services (9) and the final category are unspecified programs (10). The EDID is defined as:
𝐸𝐷𝐼𝐷 =1 2 | 𝑀 𝑀 − 𝐹 𝐹|
M and F stand for the total number of men and women in employment and the subscript i denotes the number of men and women in the field of study. The values vary between 0 and 1, with a lower value implying less segregation across fields of studies.
Segregation occupations
In the literature this variable is known as ´the sum of absolute difference in the distribution of female and male employment across occupation or sectors. It assumes that segregation implies a different distribution of women and across occupations or sectors: the less equal the distribution, the higher the level of segregation’(European Commission 2009). The occupations that are included are managers (1), professionals (2), technicians and associate professionals (3), clerical support workers (4), services and sales workers (5), skilled agricultural, forestry and fishery workers (6), craft and
19
The Karmel and MacLachlan (1988) IP index is another measurement of segregation. This index has the advantage of taking into account the female and male share of employment. However it is hard to compare the values over time, since an increase in the female share of employment tends to increase the level of
22 related trades workers (7), plant and machine operators and assemblers (8) and elementary occupations (9)20. The formula for SEGID is defined as:
𝑆𝐸𝐺𝐼𝐷 =1 2 | 𝑀 𝑀 − 𝐹 𝐹|
M and F stand for the total number of men and women in employment and the subscript i denotes the share of men and women in the specific occupation. The values vary between 0 and 1, with a lower value implying less segregation across occupations.
4.4 Control variables
The standard determinants of economic growth have been examined extensively. Following the literature the relevant determinants will be given below21. Descriptive statistics on the control variable are illustrated in table 12 and 13.
[Insert table 12 and 13]
4.4.1 Financial determinants Initial level of GDP per capita
A first determinant of economic growth is the initial level of income per capita (Barro 1991 and 1997, Sachs and Warner 1995, Harrison 1996, Easterly and Levine 1997). Such a determinant is used in either a neoclassical theory of economic growth or an endogenous growth theory. In these theories poor countries will catch up or converge to income levels of rich countries, in other words, conditional convergence is a result. It is worth noting that convergence depends on government policies on institutions, trade policies and education. According to neoclassical economists, the coefficient of initial income will be negative and it suggests that convergence to a steady state takes a long time.
Investment
According to Penn World Table, research and development (R&D) expenditures refers to “creative work undertaken on a systematic basis in order to increase the stock of knowledge, including knowledge of man, culture and society, and the use of this stock of knowledge to devise new applications”. Most of the R&D activities are carried out by companies, universities and the government. The purpose of these activities are future- and long-term oriented and frequently to create new technologies. R&D is therefore important for economic growth and necessary for
20
Women are not represented in the category army, therefore this category is being left aside. Since the ID tends to increase with the detail of the classification categories.
21
23 sustainable development (Barro 1991, Sachs and Warner 1995, Barro 1997). At the same time, more GDP per capita can stimulate growth in a country, in other words, reserve causation. The variable is measured as a percentage of the level of investment to GDP per capita, which is also shown by the following equation:
𝐿𝑜𝑔( 𝑅&𝐷
𝑃𝑃𝑃 𝑐𝑜𝑛𝑣𝑒𝑟𝑡𝑒𝑑 𝐺𝐷𝑃 𝑝𝑒𝑟 𝑐𝑎𝑝𝑖𝑡𝑎 (𝑎𝑡 2005 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 𝑝𝑟𝑖𝑐𝑒𝑠))
Regarding the inequality bias it is important to take the expenditures on R&D into account, because new technologies or knowledge can create more demand on the labor market. Thereby wages could rise, which can give women an incentive to participate in the labor force.
Trade openness
An important indicator which should not be forgotten in analyzing the effect of gender inequality is the impact of greater international integration (Barro 1991, Easterly and Levine 1997, Dollar and Kraay 2003, Alcala and Ciccone 2004, Helpman 2004, Sachs and Warner 1995). The relationship between trade openness and gender inequality may have two implications. First, a greater amount of openness could result in more export, which eventually can narrow the gap between skilled and unskilled labor. Especially for developing countries, more skilled labor creates higher relative wages, thereby creating an incentive for women to participate on the labor market. Conversely, more export can create a demand for skilled labor and then it would create a greater wedge between the wage for skilled and unskilled labor. Several studies have tried to take into account trade as a determinant of growth and the most common measure of trade openness is the following one:
𝐿𝑜𝑔( 𝐸𝑥𝑝𝑜𝑟𝑡 + 𝐼𝑚𝑝𝑜𝑟𝑡
𝑃𝑃𝑃 𝑐𝑜𝑛𝑣𝑒𝑟𝑡𝑒𝑑 𝐺𝐷𝑃 (𝑎𝑡 2005 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 𝑝𝑟𝑖𝑐𝑒𝑠))
24
4.4.2 Demographic factors Fertility rates
These days the population of Europe is ageing. Several reasons are cited as causes for an ageing population. First, a growing welfare state creates better living standards, which create a longer life expectancy. Another reason is secularism, which is the reduction of the influence of the church and religion by the secularization of the state and society. Feminism has an impact too, for example by creating access to all professions and education for women. As mentioned above and according to the literature (Klasen 1991, Barro 1991, Dollar and Gatti 1991), enrollment in education has an indirect effect on economic growth through fertility rates. Simultaneously, Sobotka et al (2010) provide evidence that fertility tends to be pro-cyclical and reacts on changes in GDP. According to the World Bank fertility rates refer to ‘the number of children that would be born to a women if she were to live to the end of her childbearing years and bear children in accordance with current age-specific fertility rates’.
Population growth
According to Bloom and Williamson (1997), fertility rates have an impact on the total population growth. Average population growth rates (Kormendi and Meguire 1985, Mankiw et al. 1992, Kelley and Schmidt 1995, Bloom and Sachs 1998) are found to have a negative effect on per capita GDP growth.
4.4.3 Institutional determinants
According the new economic growth literature, an important lesson to be learned is that institutions are important for a country’s progress. The term institutions is very broad, however, as it includes the rules of law, the openness of markets, political stability and democracy, as well as health and finance. Gender gaps are measured in different dimensions. In education and employment it might be the case that institutions for a large part determine the access opportunities of both sexes. This view is developed by institutional economic theorists. Therefore controlling the growth regression for this variable will be important. Country’s institutions are responsible for the property and civil rights of their citizens. As new rights and duties are always constructed, the traditional division of roles will erode and, as we have seen above, women will become part of the workforce. As we cannot evaluate everything in the regression, two indicators for institutions are determined and will be discussed below (Knack and Keefer 1995).
25 market. Following this research (and Barro and Lee 1994), the Gastil index22 is used here to determine the degree of democracy and political freedom in each country. An annual survey by the U.S.-based Freedom house reports the concept of these two variables, political rights and civil liberties. According to the Freedom house, political rights are the rights to participate in the political process. For civil liberties a definition is constructed that entails rights to have freedom of expression or to demonstrate and rights that provide a degree of autonomy, such as freedom of religion and school. Seven categories provide an instrument to classify countries on a scale from 1 (most rights) to 7 (fewest rights). The average of the civil and political rights determine an overall status from ‘free’ (1.0-2.5), ‘partly free’ (3.0-5.0), or ‘not free’(5.0-7.0). These average ratings are used here.
5. Descriptive statistics
This section first takes a look to get an idea of the data. Firstly, the dependent variable is analyzed over time. Secondly, the overall level of education and gaps in education and labor force participation will be discussed. Since segregation indices are new in relation to this subject, they will be discussed in more detail here. With some caution, the relationship between the dependent variable and the explanatory variables will be analyzed by using scatterplots.
5.1 Dependent variable economic growth
Table 4 represents descriptive statistics for dependent variable economic growth for the different models.
[Insert table 4]
Figure 4 shows the development of economic growth for each country over time23. The overall pattern is that countries have a positive economic growth, which lies on average around 4 or 5 percent. However, a large drop in the growth is seen from the beginning of 2007, a result from the currency crisis. This downward pattern reaches its lowest point in 2009 and thereafter the growth rate ascends again.
5.2 Explanatory variables
A summing up of the explanatory variables is defined below. Tables 5 to 13 give descriptive statistics for explanatory variables.
[Insert tables 5, 7, 8, 9, 10, 11, 12 and 13]
22
Freedom of the World publication was launched in 1973 by Raymond Gastil. Since data is not available for years before 1973 the averages for political and civil rights is based on years 1973-1975, for the first period.
23
26
5.2.2 Primary education
The main purposes of primary education are achieving basic literacy and numeracy. Laws on compulsory school attendance provide the basis for education. According to UNESCO (2012) minimum time span of compulsory education in all EU countries is seven years. Most EU countries stipulate however that children should go to school between 10 and 14 years in total. Only Estonia, Latvia, Poland and Lithuania require compulsory school attendance for between seven and nine years. So, a result of compulsory education is that for primary education nearly 23 of the 27 countries have a GER of more than 98 percent for both sexes, which is a sign of near-universal primary enrolment (table 5). Noteworthy that participation in primary education is indeed gender equal and therefore not interesting to take into account in the regressions about gender inequality.
5.2.3 Secondary education
At this level, students usually have an opportunity to choose specific programs, which can lead to different career paths. This level of education either prepares for university or college, or directly for the labor market. Secondary education is only partly compulsory and it is not directly clear if it is a consequence of this, but more variations are shown in the enrolment indicator of secondary education than in primary education (table 6). In fact, there are seven countries which favor women consistently, namely Finland, Czech Republic, Estonia, Ireland, Portugal, Sweden en Spain. In contrast, girls are disproportionately excluded from education in Austria, Bulgaria, Germany, Italy, Malta and the Netherlands (UNESCO 2012). In addition, table 13 shows that gender parity increased throughout the EU. This is shown by the GPI24 average itself, but also by the lower variance. So reaching gender parity in secondary education has become a fact.
5.2.4 Tertiary education
Finally progress in access to education has been strong for tertiary education and considerable progress has been made by the increase in the enrolment rate of females. For females the average increase in enrolment is 58 percent and for males it is merely 37 percent from 1970 till 2010. Obviously the increased enrolment of females is illustrated by the GPI of 1,33 too (table 13). Concluding, for tertiary education, the disparity has made a switch from being in favor of males to being in favor of females nowadays. Women are the great beneficiaries of tertiary enrollment and they now account for the majority of students. Clearly this is a consequence of the changed attitude of the society towards education for females. It is obvious that the traditional role of women has changed from housekeeping and taking care of the family towards a role of (part) wage-earner.
24
27
5.3 Overall level of participation
Particularly the latter half of the twentieth Century saw a rapid growth in the participation of women in Europe’s labor force. Until the early 1960s, women traditionally took care of the children and pursued domestic duties, while men were the breadwinners. Several factors account for the rapid growth and the entry of women into the labor force. First, a change took place in the social attitude towards the role of women in the household and the acceptance of their contributing to the family income. Second, the demand for female labor in the service and health sectors increased. In the past 30 years significant improvements in women’s labor participation has taken place. According to ILO (2012) labor force participation rates of both males and females has increased for the Eurozone. The largest gains are related to women, with an increase of 20.000 participants, while men only increased by 8.000 participants. The difference between the sexes becomes smaller and consequently it slowly converges to the same participation rate. However, men still have on average a higher participation rate than women (table 8).
5.4 Segregation
5.4.1 Segregation in education
Differences between the sexes can also be observed in the field in which men and women choose their studies. As figure 6 in the appendix shows men have a high participation rate in research and science (5), engineering, manufacturing and construction (6), agriculture (7) and in services (9). The proportion of female graduates is much higher in humanities and arts (3), social sciences, business and law (4), education (2) and health and welfare (8). The overall distribution for the EU of the various fields of studies is shown in figure 7, appendix. It is clear that the overall picture is almost identical to that for every European member individually (see figure 6). In addition, the index of dissimilarity is measured to evaluate the distribution of the several education fields compared to total education. Figure 6 illustrates that the index of dissimilarity is above 0,2 and below 0,5 in every country. It implies that the segregation and thereby the distribution among occupations between men and women is different. An unequal distribution means a higher level of segregation. In other words, there is a more skewed distribution in the fields of education for females than for males.
5.4.2 Segregation in occupations
