Gender gap in the labour market: a comparative view for youth and adults

Fatma Didin Sonmez and Yasemin Ozerkek

1 Introduction

There are considerable gender differences in many aspects throughout the world. Some of the prominent gender gaps occur in labour markets. Labour economists have long tried to investigate these differences. Many countries have gender-based inequality problems in their labour markets. Gender equality is an important phenomenon for an efficient labour market and economic environment to achieve sustainable development.

There exist several studies analysing the labour market outcomes such as employment and unemployment by age and/or gender in the literature (O’Higgins, 1997, 2003, 2012, 2015; Korenman and Neumark, 1997; Blau and Kahn, 2000; Blanchflower and Freeman, 2000; Breen, 2005; Pastore, 2018; Tahlin and Wes- tennan, 2020, among others). In his studies with European countries, O’Higgins (1997, 2015) and Bell and Blanchflower (2011) (with the cases of United States and United Kingdom) report that youth unemployment rates of young people are higher than those of adults and variations in youth unemployment have a close relationship with the variations in adult unemployment.¹

Besides the bulk of the literature investigating age and gender differentials, many studies focus on the gender gap in unemployment and employment (Alba- nesi and Sahin, 2013; Bicakova, 2014; Baussola, Mussida, Jenkins and Penfold, 2015; Dauth, 2016, among others) Niemi (1974) points out three factors accounting for the relatively high unemployment rate of females:

a high level of frictional unemployment because of movement in and out of the labour force; a relative lack of training, particularly specific training, and consequently a susceptibility to cyclical layoffs and unemployment; and occupational and geographic immobility, resulting in a high level of structural unemployment.

(P- 331)

In addition, Niemi (1974) mentions one other factor which lowers the female unemployment rates relative to male rates. Since the industrial distribution of the male and female labour forces differ (i.e., some industries employ mostly male workers), this leads to causing men to suffer the most from the economic recession (Niemi, 1974; Sahin, Hobiju and Song, 2009; Albanesi and Salim. 2017).

Albanesi and Sahin (2017) analyse gender differences in unemployment for a group of OECD countries in view of the female and male labour force attachment. Azmat, Giiell and Maiming (2006)’s findings show that the unemployment rates of males and females are close for some OECD countries, while especially for Mediterranean countries, the gap is large in favour of men. They also analyse the gaps in employment and unemployment flows. They point out that the rise in labour market attachment is just some part of the explanation for the changes in the gender unemployment gap. Labour market institutions, by changing the effect of human capital differences on unemployment rates, are likely to account for the gaps (Azmat, Gtiell and Manning, 2006). In a recent paper, Tahlin and Wester- man (2020) study on the causes of declining youth employment by using data for a group of European countries. They suggest that the structural change of skill in the form of skill upgrading and skill mismatch affects young people’s employment opportunities inversely.

This study includes two main strands of work. First, some facts emerged from descriptive evidence for the labour market gender gaps by age groups are delineated. Second, the empirical part analyses the sensitivity of the gender unemployment gap to the level of the unemployment rate for different age groups and investigates the relationship between the gender unemployment gap, gross domestic product (GDP) and inflation.

Considering factors such as unemployment rate, relative cohort size, employment rate, and labour force participation, descriptive part of the study aims to reveal main findings for gender-based inequalities in the labour market for a group of European countries. The study examines the gender disparities also by taking the age groups into account. Thus, the results will enable us to understand how these gender gaps differ for the youth (ages 15-24) and adults (ages 25-64). Employing amiual labour force data available in Eurostat Database (2019), descriptive analysis summarises countries’ gender gap performance.

The empirical part also highlights the fact of the unequal representation of women in different sectors. Adult and youth employment rates for different sectors including agriculture, manufacturing and services, point to gendered employment distribution in many countries. In fact, employment distribution across sectors is gender-biased; females are less represented in the manufacturing sector and more in the service sector. In the second part of the descriptive analysis, employment data for different age groups are used to reveal the key findings on the gender employment gap and the differences between youth and adult employment by sectors.

The first part of the empirical analysis explores the linkage between gender unemployment gap and the unemployment rate. To this end, the elasticity of the gender unemployment gap with respect to unemployment is estimated for the two age categories separately. The last part of the study presents models estimated by panel vector auto-regression (VAR) to illustrate the relationship between the gender unemployment gap and two basic macroeconomic variables, namely GDP per capita and inflation. These macroeconomic indicators are ultimate determinants of total demand and supply, human capital accumulation and labour market. Therefore, the unemployment rate is expected to be influenced mainly by economic activities. However, economic activities may have a disproportional effect and enlarge the gender-based inequalities in the labour market. Hence, the study aims to investigate whether the economic activities are able to generate the expected positive employment effect in terms of gender-based point of view. Besides, the model estimations for two different age groups allow comparing how this relationship differs for youth and adults.

The structure of this chapter is as follows. Section 2 discusses some gendered employment facts observed from the descriptive analysis for youth and adults. It comprises of four different subsections; unemployment, labour force participation, employment and sectoral gendered employment. Section 3 includes the gender unemployment gap sensitivity to the unemployment level. The model specification and estimation results for the relationship between the gender unemployment gap and the main macroeconomic variables are presented in Section 4. Finally, Section 5 concludes.

2 Descriptive analysis and some gendered facts in labour market

Using different age groups, descriptive evidence in this part displays gender inequalities for both youth and adults. Annual data of 27 European countries from Eurostat Database (2019) is used. Table 7.1 lists the countries included in the sample. The table also shows the country’s performance in the Global Gender Gap Index provided by World Bank (2020). Measuring gaps rather than levels, this index ranks 149 countries based on four major parameters; economic participation and opportunity, educational attainment, health and survival, and political empowerment. Thus, the Global Gender Gap Index is not an index focusing on only economic variables. Also, the economic dimension considers the participation (labour force participation) gap, the remuneration (wage, income) and advancement (occupation). Index score measures how much of the gender gap is closed. It takes a range of values from 0 to 1. The high value of index performance means greater gender equality based on the aforementioned categories in the countries.

Global Gender Gap Index is available for the period 2006-2018, except for 2017. To track the country’s progress on gender equality performance, Table 7.1 shows index score and rankings for the initial year and the last two years. Iceland ranks first in the index, followed by Norway, Sweden and Finland. These countries consistently occupy as top four in 2006, 2016 and 2018. Hungary, Slovakia, Czechia and Italy have the lowest level of scores across the European countries reported in the table. Moreover, among those listed countries, France and Slovenia are the countries making the greatest progress. France increased 58 places (from 70 in 2006 to 12 in 2018) and Slovenia raised 40 places (from 51 in 2006 to 11 in 2018) in the index. Table 7.1 reveals the fact that 11 out of 27 countries have closed

Table 7.1 Global Gender Gap Index Performance of Countries

Source: World Bank (2020)

between 67 per cent and 74 per cent of gender gaps in their countries, whilst 16 countries have closed between 75 per cent and 86 per cent.

Country’s scores and rankings in the Global Gender Gap Index allow a comparison of countries in different aspects such as economic participation and opportunity, educational attainment, health and survival and political empowerment. Thus, Table 7.1 provides a helpM list to rank countries based on their gender gap performances in different categories. However, this paxt of the study concentrates on economic issues, specifically on gender gaps in unemployment, employment and labour force participation. The selected countries in the analysis have some similarities and also disparities based on the labour market issues discussed in the following subsections.

2.1 Unemployment

Figure 7.1 shows the evolution of youth and adult unemployment rates for 27 European countries. The most obvious striking characteristic of the figure is that youth unemployment rates are higher than adult unemployment rates in all countries during 1995-2018.² Italy, Croatia and Spain among the others, are the countries that suffer the most from very high youth unemployment. Figure 7.1 also illustrates that both rates generally move together. The correlation between these rates (0.87) is pretty high. The lowest correlations are in Sweden (0.39) and Austria (0.37).

In some countries, for low levels of unemployment rates, the gap between youth and adults is narrow. One observes this pattern, especially in Switzerland, Germany and Austria. The gap between adult and youth unemployment rates are remarkably larger in Poland, Croatia, Italy and Spain.

The response of youth unemployment to economic fluctuations is remarkably higher than that of adults. The youth rates tise more in downturns and recovering more rapidly during expansions (O’Higgins, 1997). This observation is true across several countries. As is seen in Figure 7.1, the unemployment rates appear to increase in response to the 2008 economic crisis, particularly affecting youth rates more than do adult rates. In Bulgaria, Croatia, Czechia, Denmark, Estonia, Hungary, Iceland, Ireland, Italy, Larina, Lithuania, Slovakia, Slovenia, Portugal, Poland, Spain and the United Kingdom, the gap between adult and youth unemployment enlarges due to the economic crisis.

Figure 7.2 provides the coefficient of variation³ showing the extent of unemployment variability across the countries for the period between 1995 and 2018. The lower the coefficient of variation, the smaller the unemployment dispersion around the mean for a specific year. For both adults and the young, there is a sharp increase in female and male unemployment variations across countries from 1999 through 2001. Starting in 2002, both female and male unemployment variations decline significantly, before following an increasing trend. As is depicted in the figure, youth and adult unemployment variations follow a similar pattern.

However, it should be noted that variation in adult male unemployment is decreasing while the variation in adult female unemployment is increasing after 2010. In recent years, unemployment variation across countries is relatively high for females and the young. Figure 7.2 also indicates that since the effect of the global economic crisis on unemployment varies significantly across countries, the coefficient of variations during economic downturn periods display higher values.

Based on the different trends in the coefficient of variations over the time period, the whole period is separated into specific time intervals as shown by the vertical lines (Figure 7.2). Table 7.2 highlights the other important aspects of unemployment over three different time periods; 1995-2001, 2002-2008 and 2009-2018. Table 7.2 presents the average unemployment rates of youth and adults by gender. Compared to 1995-2001 numbers, the average rates of unemployment declines in the 2002-2008 period, and increases even more in the 2009-2018 period. This is the case for all young people and adults regardless of gender. Figure 7.1 and Table 7.2 together reveal the fact that during the period where the dispersion of

132 Fatma Didin Somnez and Yasemin Ozerkek

unemployment across countries is falling, unemployment rates are declining on average as well.

It is also prominent that youth rates and adult rates are moving apart. Female and male youth unemployment rates are higher than their counterparts for adults.

Figure 7.1 (Continued)

Figure 7.2 Variations in Unemployment Across Countries (1995-2018)

Source: Authors’ own calculations

The unemployment rates for females and males are synchronised most of the time. The intensity of this co-movement increases from 1995-2001 to 2002-2008 and from 2002-2008 to 2009-2018.⁴

In countries like Italy, Croatia, Slovakia and Spain, youth unemployment rates are persistently high. Considering three different time spans, Bulgaria and Finland make remarkable progress in reducing youth unemployment that reached very high levels in the first period, whilst Italy and Spain show a retrograde movement.

Source: Eurostat Database (2019) * Not available

There are many countries with high unemployment rates for the young over the time peiiods mentioned in Table 7.2. The shares of the countries having 20 per cent and over youth unemployment rates are 30 per cent, 26 per cent, and 52 percent in consequent time periods.⁵ In the first period, 63 per cent of these countries with the youth unemployment rate of 20 per cent and over, have higher female rates than male rates. For the second and the third periods, the corresponding rates are 57 per cent and 36 per cent. These results indicate that in these countries, the disadvantage of young females in the form of relatively high rates compared with males is gradually decreasing.

The link between gender unemployment gap, defined as the difference between male and female unemployment rate, and real GDP per capita for youth and adults is depicted in Figure 7.3. In the figure, the average values of the variables are plotted for the period 1995-2018. The negative relationship for the variables is observed for both age groups. In addition, for youth, unemployment gaps are more dispersed in the scatter diagram compared to those of adults. Majority of the countries are positioned in the same way, suggesting that the gender unemployment gaps for youth and adults are showing a similar pattern. Unemployment gaps vary between -6 per cent and 4.7 per cent for youth and -5.5 per cent and 3 per cent for adults. The number of countries with negative gender unemployment gaps (i.e. female unemployment is higher than male unemployment) is 16 for adults and 9 for youth out of 27 countries. Although female adults have relatively higher unemployment rates than male adults in many countries, the gap is not reaching that much high values as in the case of youth.

Figure 7.4 provides evidence on the relationship between the average gender unemployment gap and relative cohort size for youth and adults. Relative cohort size is measured as the ratio of the female labour force by the male labour force. Reference lines show the average level of relative cohort size and zero unemployment gap for youth and adults. All countries have relative cohort size smaller than 1, suggesting that the male labour force is greater than the female labour force.

Figure 7.4 Unemployment Gap Versus Relative Cohort Size (1995-2018) Source: Eurostat Database (2019)

with only exceptions of Lithuania and Latvia for adults. The adult female labour force is slightly above the adult male labour force in these countries. Estonia has also a ratio closed to 1. On the contrary, the young female labour force is quite lower relative to the young male labour force in Lithuania, Latvia and Estonia. Italy, Spain, Luxemburg and Ireland have the lowest relative cohort size for adults. As illustrated in Figure 7.4, adult unemployment gap values are mostly concentrated around the intersection of two reference lines, whereas for young people, the values of gender unemployment gap and relative cohort size are widely spread.

2.2 Labour force participation

Figure 7.5 illustrates the average level of female and male labour force participation (LFP) rates by age for the peiiod 2014-2018. The 45-degree line is used to investigate whether a gender-biased distribution exists for LFP. It is a striking fact that male LFP rates are considerably higher than those of female for adults. Youth LFP rates display a similar gendered distribution at relatively low levels while the difference between male and female rates shrinks at high rates of LFP for the young. Adult LFP rates range from 78 per cent (Croatia) to 94 per cent for male (Iceland) and from 61 per cent (Italy) to 86 per cent (Iceland) for female, respectively. Figure 7.5 reveals more dispersion in LFP for youth than adults. Among the reported European countries, the maximum level of youth LFP is observed in Iceland (82 per cent for female and 80 per cent for male), followed by Switzerland (almost 68 per cent for female and male) and the Netherlands (69 per cent for female and 67 per cent for male).

Figure 7.6 depicts the average gender gaps in unemployment and LFP by age for the period 2014-2018. The fitted line with a negative slope portrays a negative relationship between the gender unemployment gap and gender LFP gap for both youth and adults. Generally, higher values of LFP gaps tend to be associated with

Figure 7.5 Adult and Youth LFP Rates (%) by Country (2014-2018) Source: Eurostat Database (2019)

Figure 7.6 Gender Unemployment Gap and Gender LFP Gap (%) by Country (2014-2018) Source: Eurostat Database (2019)

lower levels of gender unemployment gap. The figure implies that notably, all countries display a positive adult gender gap in LFP,⁶ meaning that adult female LFP rate is less than male in the European countries plotted. Most of the countries have also positive gender LFP gap for youth while only 7 out of 27 countries, namely Switzerland, Netherlands, Sweden, Denmark, Iceland, Norway and Finland have a negative one.

2.3 Employment

The gender employment gap is defined as the difference between the employment rates of males and females. The employment rate is computed by dividing the number of persons in employment by the total population of the same age group.

Eurostat calculates the gender employment gap for the 20-64 age group. In this study, the rates are computed for 15-24 and 25-64 age groups by using the same methodology as Eurostat used.

The gender gap in employment rates for youth and adults are plotted for the last five-year averages in Figure 7.7. Reference lines are representing average values. The average gender employment gap is positive for adults in all countries, indicating that the adult female employment rate is smaller than the males. As for the case of young people, gender gaps are negative for seven countries (Norway, Finland, Sweden, Denmark, Iceland, the Netherlands and Czechia). Put differently, females are more advantageous than males with their greater employment rates in these countries. Indeed, gender gaps have the smallest values in Denmark, Iceland, Sweden and Finland. Although the employment gender gap for adults has very low levels in Lithuania and Latvia, young females have lower employment rates than young males. On the other hand, both the youth and adult gender employment gaps are apparently high in Italy, Hungary, Poland and Slovakia. In addition, Ireland, Spain, the United Kingdom, Czechia and Luxemburg have relatively large (between 10 per cent and 15 per cent) gender employment gaps for adults, but corresponding rates for young people is quite low (even negative for some).

2.4 Sectoral employment

This part of the study reveals the main findings regarding employment distribution by sex, age, and economic activity for 27 European countries. Sectoral employment data for agriculture, manufacturing and sendee provided by Eurostat database is used to scrutinise whether there is a gender-biased employment distribution for adults and the young. Figure 7.8 shows the average male and female

employment shares⁷ by age groups in different sectors for the period 2014-2018. It also includes a 45-degree solid line to point out a gendered employment distribution across sectors. The left part of the line demonstrates the countries having a higher level of male employment share than that of the female. The right part of the line shows the countries with a higher level of female employment share than a male has for a specific sector.

Figure 7.8 shows that agricultural employment shares for both male and female are very low compared with employment shares in the manufacturing atrd service sectors in the countries. Approximately, on average only 2 per cent of female employees and 5 per cent of male employees are working in the agriculture sector. The maximum levels of employment shares are 12 per cent for males (Lithuania and Poland) and 9 per cent for females (Slovenia and Poland). Also, there is a positive strong correlation between youth and adult male employment share in agriculture while this correlation is moderate for females.⁸

The average female employment share in the manufacturing sector is roughly half that for male. On average, 20 per cent of total male employees and 10 per cent of total female employees are working in this sector. Czechia has the maximum shares of youth and adult employment in the manufacturing sector, (i.e. 45 per cent and 31 per cent for males, 25 per cent and 21 per cent for females, respectively). For both females and males, there is a strong relationship between the proportions of the young and adults employed in the manufacturing sector.⁹

In the service sector, both female and male employment proportions in total employment are remarkably high in comparison to the other sectors (Figure 7.8). Moreover, employment proportions vary considerably for adults and youth in the service sector, although these rates change relatively less in the manufacturing and agricultural sectors. Male employment shares range from 38 per cent (Slovakia) to 76 per cent (Denmark) for youth and from 45 per cent (Poland) to 73 per cent (Luxembourg) for adults. Corresponding shares for females range from 51 percent (Bulgaria) to 93 per cent (United Kingdom) for the young and from 73 percent (Bulgaria) to almost 92 per cent (Norway and Sweden) for adults. The link between adult and youth employment is not as strong in the service sector as it is in the manufacturing sector.¹⁰

It is apparently observed that regardless of the age group, agriculture and manufacturing sectors mainly employ males while service sector employs relatively more females. Almost every covrntry analysed in this study, has higher employment shares of males than females in manufacturing atrd agriculture, with a dispropor- tionally large share of female workers in the service sector. Briefly, women are more likely than men to be represented in the service sector atrd among the young and adults similar disparities exist.

3 The gender unemployment gap sensitivity to unemployment level

It is generally accepted that if aggregate unemployment rates are high, both females and males are less likely to be employed. Art increase in the unemployment rate for a specific age group may create high unemployment levels for both male and females but the relative effect in terms of gendered unemployment may differ. In order to examine the existence of such a disproportional effect, this part of the study focuses on the sensitivity of the gender gap in unemployment to the level of unemployment for youth and adults.

There are several studies investigating the elasticity of youth unemployment with respect to relative cohort size and adult unemployment.¹¹ This study contributes to the literature by jointly employing the variables with gender and age aspects. The models to gauge the elasticity of gender unemployment gap with respect to the unemployment rate are as follows.

where youth_un, adultjm, youthungap and adult_ungap are the unemployment rates and the gender unemployment gaps (calculated as the male unemployment rate minus the female unemployment rate) for youth and adults. The valuables of cohoit_youth and cohort adult stand for the relative cohort sizes, defined as the female labour force divided by the male labour force for youth and adults, correspondingly. Ln indicates the natural logarithm.

The preceding two models are estimated by the fixed effect estimation method, which incorporates the time-invariant factors. Table 7.3 reports the estimated results of the models. An increase in the unemployment rates shows similar effects on gender unemployment gap for both youth and adults, suggesting that a 1 per cent increase in the unemployment rate will raise gender unemployment gap by around 0.83 per cent. An elasticity of unemployment gender gap with respect to the unemployment rate of less than one implies that as unemployment rates increase, unemployment gender gap also increases but less than proportionately. This implies that change in male unemployment is more than female. It is a well-known fact that some sectors have more male workers than female workers (e.g. the manufacturing sector). If male-dominated sectors are more affected by unfavourable economic conditions (say, in terms of high unemployment), this situation is expected to generate a disproportionate change against male workers. Likewise, in the presence of favourable economic conditions with declining unemployment rates, new job opportunities seem to benefit males in those gender- biased sectors more.

An increase in the relative cohort size of 1 per cent for young people will raise the youth gender unemployment gap by around 2.2 per cent. In other words, as the ratio of the young female labour force to the young male labour force increases, young males become disproportionally more unemployed than females. However,

Table 7.3 The Elasticity of Gender Unemployment Gap with Respect to Unemployment

Source: Authors’ own calculations. Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1

the corresponding elasticity coefficient for the adults in Model 1 is not significant. That is, the relative cohort size does not have significant implications on the gender unemployment gap for adults. Thus, the relative (male and female) youth unemployment rates of young people are sensitive to the ratio of female to the male labour force. In addition, although the youth unemployment rate has a significant effect on the gender unemployment gap in youth, relative cohort size has a more important influence.

4 Gender unemployment gap, GDP and inflation

From an empirical point of view, the linkage between unemployment, GDP and inflation is explained by Okun’s Law and Phillips Curve. There are enormous empirical studies checking the validity of Okun’s Law which states a negative relationship between changes in unemployment and real output. It is possible to say that there is no unique result in the literature. Results are highly sensitive to country-specific factors, time periods and the samples selected but generally consistent with Okun’s Law for developed countries.¹² Some of those studies focused on European countries. Hutengs and Stadtmann (2013) explore Okun’s Law relation for a sample of European countries for different age groups. They find that young unemployment rate is responding more to economic growth.¹³ Considering both age and gender, Dunsch (2017) iirdicates that unemployment sensitivity to business cycles is higher for youth but gender does not matter in Central and Eastern European countries. By analysing a panel of European countries, Brincikova and Danno (2015) show that the effect of economic growth on unemployment is higher for males and sensitivity of gender-specific unemployment is higher for the countries with a lower level of economic activities. Hence, it is worth studying both age and gender aspects in the analysis.

The unemployment rate is also linked to inflation by the Phillips Curve which stresses the existence of a trade-off relationship between inflation and short term unemployment fluctuations. A low level of unemployment is expected as inflation increases because of expansions in economic activities which provide new job opportunities. Thus, unemployment, inflation and GDP are highly interdependent variables.

Our study contributes to the literature by considering the bi-directional relationship between gender unemployment gap, real GDP and inflation. Comparing with existing literature, this study analyses the link between these variables for a group of European countries, by using a different model specification which jointly takes age and gender into account.

4.1 Model specification

As is stated by Okun’s Law and Phillips Curve, inflation and GDP per capita are closely linked to unemployment. Focusing on disproportional gendered unemployment, this part of the study examines the interaction between gender unemployment gap and two basic macroeconomic variables, namely inflation and GDP per capita. In the analysis, annual data of inflation measured by consumer price index (CPI), real GDP per capita, male and female unemployment rates for 27 European countries for the period 1995-2018 are used.¹⁴ Unemployment gap is calculated as the difference between male and female unemployment rates for adults aged 25 to 64 and the young aged 15 to 24.

Because of the possibility of multidirectional interactions of these variables, panel vector auto-regression (VAR) model is used. This methodology including a system of equations allows us to enter all variables as endogenous. It also derives the impulse-response functions representing the dynamic response of an endogenous variable to a shock in another variable in the model.

The following regression model was used for VAR analysis including three endogenous variables:

where un gap is the gender unemployment gap computed as the difference between male and female unemployment rates, GDP is the log real GDP per capita, inf is the inflation measured by CPI, wi and e_j are panel fixed-effects and idiosyncratic

Table 7.4 MBIC, MAIC and MQIC Information Criteria

Source: Authors’ own calculations

Figure 7.9 Roots of the Companion Matrix Source: Authors’ own calculations

error terms, respectively. The preceding equations are estimated for both adult and youth unemployment in Model 1 and Model 2 correspondingly.

For the analysis, the panel VAR method in the study of Abrigo and Love (2016) is employed.¹⁵ Lag order selection is made by using MBIC, MAIC and MQIC¹⁶ information criteria presented in Table 7.4 (Andrews and Lu, 2001). The appropriate lag length which gives the smallest value of information criteria is determined as 1 for both Model 1 and Model 2.

The eigenvalues of three different endogenous variables in the model are calculated. Figure 7.9 shows that all the eigenvalues are smaller than one in modulus and lie inside the unit circle. In other words, all moduli of the companion matrix are less than one. Thus, the constructed panel VAR models satisfy the necessary and sufficient conditions for stability (Abrigo and Love, 2016).

4.2 Results

The estimation results are reported in Table 7.5. All variables in the models are positively affected by their lagged values for both youth and adults. There is highly

146 Fatma Didin Sonmez and Yasemin Ozerkek Table 7.5 Estimation Results

Source: Authors’ own calculations. Standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1

significant bi-directional causality between adult gender unemployment gap and real GDP per capita (Model 1). However, for the young, there exists only one-way causality from the gender unemployment gap to real GDP per capita (Model 2). Both adult and youth gender gaps in unemployment affect real GDP per capita inversely, whereas real GDP per capita has a positive impact on the gender unemployment gap for adults but not for youth. Inflation is found to be insignificant in determining adult and youth gender unemployment gaps. Table 7.5 also signifies a negative bi-directional causality between real GDP and inflation for Model 1 and one-way causality from inflation to real GDP per capita for Model 2.

Stability of estimated panel VAR analysis allows driving impulse-response functions. Figure 7.10 illustrates the impulse-response functions and confidence intervals (i.e. the shadowed area) calculated by using Monte Carlo simulations with 500 iterations. Each function in the figure demonstrates the reaction of an endogenous variable which is normalised by one standard deviation to one standard deviation exogenous shock in another variable in the dynamic systems of equations over the step horizon specified as ten periods. Since there is no statistically significant relationship between inflation and the gender unemployment gap, related impulse-response functions are not reported.

The functions estimated for Model 1 and Model 2 imply that a positive shock to real GDP per capita has a statistically significant positive impact on the adult gender unemployment gap as well as a negative effect on inflation. However, this exogenous shock does not have any statistically significant effect on the youth gender unemployment gap.¹⁹ The maximum effect is experienced in two periods and it lasts in seven periods for the adult gender unemployment gap. Moreover, a positive exogenous shock in youth and the adult gender unemployment gap creates a negative impact on real GDP per capita. This effect peaks in two periods and lasts approximately in five periods. The results also suggest that adult gender unemployment gap has a larger impact on real GDP per capita in comparison to the youth gender unemployment gap.

Figure 7.10 Impulse-Response Functions Source: Authors’ own calculations

5 Conclusion

This study concentrates on gender gaps which exist in the labour market in a comparative view for different ages and the relationship among gender unemployment gap, GDP and inflation. Descriptive analysis part presents some gendered labour market facts observed, specifically for unemployment, employment and labour force participation.

The most obvious fact is that youth unemployment rates are higher than adult unemployment rates in all countries and some of those countries are suffering persistently very high youth unemployment. The shares of the countries having 20 per cent and over youth unemployment rates are increasing over time. Although female adults have relatively higher unemployment rates than male adults in many countries, the gap is not reaching that much high values as in the case of youth.

Moreover, the response of youth unemployment to economic fluctuations is distinctly higher than that of adults. In most of the European countries, the gap between adult and youth unemployment enlarges due to the economic crisis. Since the effect of the global economic crisis on unemployment varies significantly, greater unemployment variations exist across countries during economic downturn pexiods. On the contrary, when the unemployment rates are declining, the variations in unemployment across countries are falling as well. Also, unemployment variation across countries is relatively high for females and the young for the last five years.

The male labour force is greater than the female labour force in almost all countries. It is another important fact that adult male LFP rates are considerably higher than those of female. Youth LFP rates display a similar gender-biased distribution across countries at relatively low levels, while the difference between male and female rates tends to close at high rates of LFP for the young. It is also observed that higher LFP gaps mostly are associated with lower levels of gender unemployment gap for both youth and adults.

In addition, all countries have positive employment gaps indicating smaller adult female employment rate than male employment for adults. Only seven exceptions out of 27 with negative gender gaps are observed for youth employment. Sectoral employment data demonstrate very' low employment shares in agriculture compared with manufacturing and service sectors. It is obviously observed that males are mostly employed in the manufacturing sector, whilst females are employed in the service sector regardless of age group. Thus for both youth and adult similar gender-biased sectoral employment distribution exist.

Additionally, an empirical investigation of unemployment gap sensitivity to changes in unemployment level finds an inelastic unemployment gap with a positive sign. This implies that change in unemployment rate enlarges gap but less than proportionately and vaiies male unemployment more.

Finally, the relationship between unemployment gaps for different age groups and two main macroeconomic variables is analysed. Panel VAR models are employed to investigate whether multidirectional interactions exist among those variables. The estimation results are different for young and adult labour force. Highly significant bi-directional causality between the gender unemployment gap and real GDP per capita is found for adults. On the other hand, there is only oneway causality from the gender unemployment gap to real GDP per capita for youth.

It is important to highlight that regardless of age group, a greater level of gender unemployment gap is associated with a lower level of GDP per capita. This implies that if relatively more males tend to be unemployed then GDP per capita declines. This result is not surprising considering the gender employment gap, labour force participation gap and gender-biased distribution of sectoral employment. Relatively smaller female employment and labour force participation compared to males in most countries suggests relatively less active female labour force in the labour market. The economy having male-dominated labour market contracts if men become more unemployed. Moreover, since males are mostly employed in the manufacturing sector which is a key sector to support the economy, relatively high male unemployment may indicate a lower level of GDR

The results of the analysis also suggest a positive significant effect of real GDP per capita on the gender unemployment gap for adults, but no significant effect for youth. This indicates that if the economy is expanding, the adult male unemployment rate decreases less than adult female unemployment does (assuming that real GDP and unemployment are inversely related). Also, lower GDP per capita is associated with a relatively larger increase in adult female unemployment. As a result, the change in real GDP per capita creates a gender-biased effect on adult unemployment, thereby affecting female unemployment rates more.

Notes

• 1 O’Higgins (1997) explains the reasons of these facts based on aggregate demand, size of the youth labour force and wages.
• 2 The only exception is for some years before 2002 in Germany.
• 3 Coefficient of variation is calculated as the standard deviation of unemployment rate across countries divided by the average unemployment rate for a given year.
• 4 The correlation coefficients between average young male and female unemployment rates for all countries in Table 7.2 are 0.90, 0.92 and 0.96 for 1995-2001, 2002-2008 and 2009-2018, respectively. Corresponding rates for adults are 0.80, 0.88, 0.92.
• 5 The countries having 20 per cent and over youth unemployment rates are as follows: 1995-2001 period: Bulgaria, Finland, France, Italy, Lithuania, Poland, Slovakia, Spam: 2002-2008 period: Bulgaria, Croatia, Finland, Italy, Poland, Slovakia, Spain; 2009- 2018 period: Belgium, Bulgaria, Croatia, France, Hungary, Ireland, Italy, Latvia, Lithuania, Poland, Portugal, Slovakia, Spain, Sweden.
• 6 LFP gap equals to male LFP rate minus female LFP rate.
• 7 Female employment share is calculated as the ratio of female employment in a specific sector to total female employment, and male employment share is calculated as the ratio of male employment in a specific sector to total male employment.
• 8 The correlation coefficients between youth and adults are 0.84 for males and 0.53 females in the agricultural sector.
• 9 The correlation coefficients between the young and adults are 0.91 for males and 0.93 females in the manufacturing sector.
• 10 The correlation coefficients between the young and adults are 0.75 for males and 0.72 females in the service sector.
• 11 See Korenman and Neumark (1997) and O’Higgins (1997, 2001, 2003).
• 12 Penman, Stephan and Tavera (2015) covers the related literature testing Okun’s Law for different groups of countries in its meta analysis.
• 13 Baneiji, Saksonovs, Lin and Blavy (2014) also present a statistical analysis which points out three times more sensitive youth unemployment rate in comparison to adult unemployment rate.
• 14 In order to detect whether variables have unit roots, panel unit root tests of ADF-Fisher (Maddala and Wu, 1999), IPS (Im, Pesaran and Shin, 2003) and LLC (Levin, Lin and Chu, 2002) are performed. CPI and gender unemployment gap (both for youth and adults) exhibit stationary characteristics in levels. On the other hand, GDP per capita has a unit root, therefore it is detrended by applying Hodrick-Prescott filter.
• 15 Stata 14 statistical package program is used in the analysis.
• 16 MAIC: Modified value of Akaike information criteria (Akaike, 1969); MBIC: Modified value of Bayesian information criteria (Schwarz, 1978; Rissanen, 1978; Akaike, 1977); MQIC: Modified value of Hannan-Quinn information criteria (HQIC) (Hannan and Quinn, 1979).
• 17 To test overidentifying restriction, Hansen’s J test is used and p-value is 0.328 for Model 1 (Hansen, 1982).
• 18 Hansen’s J test p-value is 0.163 for Model 2.
• 19 Since it is not statistically significant, impulse-response function is not reported.

References

Abrigo, M. and Love, I. (2016). Estimation of panel vector autoregression in Stata. Stata Journal, 16 (3), 778-804.

Akaike, H. (1969). Fitting autoregressive models for prediction. Annals of the Institute of Statistical Mathematics, 21, 243-247.

Akaike, H. (1977). On entropy maximization principle. In P.R. Rrishnaiah (eds.), Applications of Statistics, Amsterdam: North-Holland, 27-41.

Albanesi, S. and Sabin, A. (2013). The gender unemployment gap. Federal Reserve Bank of New York Staff Report No. 613.

Albanesi, S. and Sahin, A. (2017). The gender unemployment gap. NBER Working Paper No. 23743.

Andrews, D.W.K. and Lu, B. (2001). Consistent model and moment selection procedures for GMM estimation with application to dynamic panel data models. Journal of Econometrics, 101 (1), 123-164.

Azmat, G., Gttell, M., and Maiming, A. (2006). Gender gaps in unemployment rates in OECD countries. Journal of Labor Economics, 24 (1), 1-37.

Banerji, A., Saksonovs, S., Lin, H., and Blavy, R. (2014). Youth unemployment in advanced economies in Europe: Searching for solutions. IMF Staff Discussion Note 14/11. Washington: International Monetary Fund.

Baussola, M., Mussida, C., Jenkins, J., and Penfold, M. (2015). Determinants of the gender unemployment gap in Italy and the United Kingdom: A comparative investigation. International Labour Review, 154 (4), 537-562.

Blanchflower, D.G. and Freeman, R.B. (2000). The declining economic status of young workers in OECD countries. In D.G. Blanchflower and R.B. Freeman (eds.), Youth Employment and Joblessness in Advanced Countries, Chicago: Chicago University Press/NBER. 19-55.

Blau, F.D. and Kahn, L. (2000). Gender and youth employment outcomes (The United States and West Germany, 1984-1991). In D.G. Blanchflower and R.B. Freeman (eds.), Youth Employment and Joblessness in Advanced Countries, Chicago: Chicago University Press NBER, 107-168.

Bell, D. and Blanchflower, D. (2011). Young people and the great recession. OxfordReview of Economic Policy, 27, 241-267.

Bicakova, A. (2014). Selection into selection into labor force and gender unemployment gaps. CERGE-EI Discussion Paper Series, 513.

Breen, R. (2005). Explaining cros-national variation in youth unemployment. European Sociological Review, 21 (2), 125-134.

Brincikova, Z. and Danno, L. (2015). The impact of economic growth on gender specific unemployment in the EU. Scientific Annals of the "Alexandra loan Cuza ", University of lap Economic Sciences, 62, 383-390.

Dauth, C. (2016). Gender gaps of the unemployed-what drives diverging labor market outcomes? IAB Discussion Paper No. 27.

Dunsch, S. (2017). Age-and gender-specific unemployment and Okun’s law in CEE countries. Eastern European Economics, 55, 377-393.

Eurostat Database (2019). Retrieved January 10,2020, from https://ec.europa.eu/eurostat/ data/database

Hannan, E.J. and Quinn, B.G. (1979). The determination of the order of an autoregression. Journal of the Royal Statistical Society, Series В 41,190-195.

Hansen, L.P. (1982). Large sample properties of generalized method of moments estimators. Econometrica, 50 (4), 1029-1054.

Hutengs, O. and Stadtmann, G. (2013). Age effects in Okun’s law within the Eurozone. Applied Economics Letters, 20 (9), 821-825.

Im, K.S., Pesaran, M.H., and Shin, Y. (2003). Testing for unit roots in heterogeneous panels. Journal of Econometrics, 115 (1), 53-74.

Korenman, S. andNeumark, D. (1997). Cohort crowding and youth labor markets: A cross- national analysis. NBER Working Paper No. 6031. Cambridge, MA.

Levin, A., Lin, C.F., and Chu, C.S.J. (2002). Unit root test in panel data: Asymptotic and finite sample properties. Journal of Econometrics, 108 (1), 1-24.

Maddala, G.S. and Wu, S. (1999). A comparative study of unit root tests with panel data and a new simple test. Oxford Bulletin of Economics and Statistics, 631-652.

Niemi, B. (1974). The female-male differential in unemployment rates. Industrial and Labor Relations Review, 27, 331-350.

O’Higgins, N. (1997). The challenge of youth unemployment. Employment and Training Papers No. 7. Geneva: Employment and Training Department, ILO.

O’Higgins, N. (2001). Youth Unemployment and Employment Policy: A Global Perspective. Geneva: ILO.

O’Higgins, N. (2003). Trends in the youth labour market in developing and transition countries. World Bank Social Protection Discussion Paper Series No. 0321.

O’Higgins, N. (2012). This time it’s different? Youth labour markets during ‘The Great Recession ’. IZA DP No. 6434.

O’Higgins.N. (2015). Youth unemployment. IZA Policy Paper No. 103.

Pastore, F. (2018). Why is youth unemployment so high and different across countries? IZA World of Labor, 420.

Pennan, R., Stephan, G., and Tavera, C. (2015). Okun’s law: A meta analysis. The Manchester School, 83(1), 101-126.

Rissanen, J. (1978). Modeling by shortest data description. Automatica, 14,465-471.

Sabin. A., Hobijn, B., and Song, J. (2009). The unemployment gender gap during the current recession. Current Issues in Economics and Finance, 16(2), 1-7.

Schwarz, G. (1978). Estimating the dimension of a model. Annals of Statistics, 6, 461-464.

Tahlin, M. and Westerman, J. (2020). Youth employment decline and the structural change of skill. European Societies, 22 (1), 47-76.

World Bank (2020). Overall Global Gender Gap Index. Retrieved January 10, 2020, from https://tcdata360.worldbank.org/indicators/af52ebe9?country=BRA&indicator=27959 &viz=line_chart&years=2006,2018