Gender Inequality in Indonesia

October 2015

A report prepared in a collaboration between the Australian-Indonesian Partnership for Economic Governance (AIPEG), the Australian Department of Foreign Affairs and Trade (DFAT) and Monash University’s Centre for Development Economics and Sustainability (CDES). 1

1

Executive Summary

This report presents a review of existing papers on the state of gender inequality in Indonesia. It also includes results of quantitative research on gender disparities in Indonesia in terms of the drivers of female labour force participation and gender wage gaps and tracks their progress over time. The Global Gender Gap Report 2 (2014) prepared by the World Economic Forum identifies inequality in economic participation and opportunity for women as the most significant gender inequality challenge for Indonesia. Economic participation is thus the main focus of this study. Summary of Existing Literature: Educational Inequality We start by examining educational inequalities. Educational attainment provides a path to better compensated and more fulfilling employment and hence educational inequality can be a root cause of observed inequality in economic opportunities later in life. As will be seen below, there is little in the way of gender education gaps amongst younger cohorts in Indonesia. Indonesian women are on average slightly better educated than Indonesian men. However, the overall educational performance of Indonesians is low. Labour Market Outcomes Female labour force participation is significantly lower than men’s in Indonesia and low relative to countries at a comparable stage of development. This reflects gender differences in family roles, child-caring and also cultural norms about women’s role. Although women’s labour force participation rates are lower than men’s, they also experience higher levels of unemployment and underemployment. Women are over-represented in the informal sector and more likely to be unpaid workers. Cultural norms also play a role in the strong gender segregation of industries and occupations, with women being concentrated in lower paying roles. Indonesia has a large gender wage gap with women being paid around 30 % less than a similarly qualified man. International migration for work has provided valuable earning opportunities for low skilled Indonesian women who struggle to find adequately remunerated work in their home country. Women are over-represented as migrant workers, working mainly as domestic workers. Although international migration contributes to family incomes, low skill women are often exploited in the destination countries. Entrepreneurship and Finance While women are concentrated in the informal sector, often in self-employment, they are also under-represented as entrepreneurs. This is often attributed to their difficulty accessing financial resources which limits their ability to develop businesses and exercise their entrepreneurial abilities. The evidence on this is however mixed and further research in this area is desirable

2

Infrastructure Inadequate transport infrastructure and services are additional barriers to women’s full economic participation. Transport – both urban public transport and rural roads – are important in providing women with employment opportunities. Research is only beginning to shine light on gender biases in infrastructure provision which often is designed to meet the needs of men rather than women. Reliable and efficient transport is particularly important for women as they are often juggling household responsibilities with employment which can make it difficult to work at a distance from home. Efficient transport reduces travel times and can make it feasible for a woman to work at a greater physical distance from home – hence opening up access to jobs and markets for products. Health Gender gaps in basic health indicators such as infant and child mortality and morbidity and utilisation of health services, although observed in many developing countries, are not apparent in Indonesia. Life expectancy for Indonesian women is 5 years in excess of that for Indonesian men. Nevertheless women’s reproductive role puts them at risk. Maternal mortality rates, although declining, are higher than in comparable countries. Access to contraception by unmarried women is very limited. Marriage at a young age is common in some areas. Women’s limited earnings capacity can limit their access to health services. There is some indication that women are more prone to mental illness than men. Institutions and Laws Gender equality is hampered in many instances by laws and institutions that deny women equal property rights, acknowledgement as a household head, and access to work. Women are also under-represented politically, making it harder for their voices to be heard. Results of the Quantitative Research: Female Labour Force Participation Female labour force participation in Indonesia has remained relatively constant from 1996 to 2013 even in the face of dramatic economic change. Our analysis however shows that once you control for individual, household and village characteristics, there are signs that the underlying propensity for women to participate in the labour force has been increasing, particularly in urban areas. This is an interesting result and is consistent with changes in societal attitudes towards females in the labour market. Offsetting this secular increase in women’s labour force participation has been decreasing participation as a result of the lesser importance of agriculture. If this trend continues then we may see female labour force participation increase as the older cohorts exit the labour market. The main drivers of female labour force participation (cohort and age effects aside) are marital status, the number of children aged between 0 and 2 years of age in the household, educational attainment (in particular at the upper-secondary and tertiary levels) and the village industrial structure (with agriculture and manufacturing being female-friendly industries, although still male dominated). That marital status and the presence of young children have such a large negative impact on female labour force participation, particularly relative to other countries, suggests that policies targeted at providing some form of child-care for women with young children may be effective. Policies ensuring that

3

women have access to the higher levels of education, particularly in rural areas where educational attainment is lower, may also boost women’s participation. That the cohort analysis finds that the underlying propensity for women to participate in the labour markets is increasing is promising. However, the ongoing movement of the Indonesian economy away from the agricultural sector, given the importance of the agricultural sector to female employment, will continue to at least partially offset this effect. Policies designed to provide women with access to employment in non-traditional industrial sectors, for example, through the provision of subsidised vocational education and/or campaigns that provide and promote opportunities for women in these sectors, are also worthy of attention. Gender Wage gap There is significant wage discrimination against women in Indonesia. The raw wage gap averages 41% and only a small proportion of the gender wage gap can be explained by differences in productive characteristics. This is true in both the formal and informal sectors and at most points across the distribution of wages. Differences in years of experience between men and women explain some of the difference in wages as women have less experience on average due to career interruptions associated with child-rearing. Industrial segregation by gender explains a large portion of the wage gap – women work predominantly in female-dominated industries like services and trade with men being concentrated in “male” industries like transport and mining. Women’s higher educational attainment works to reduce the wage gap. Overall, the explained proportion of the gap is not larger than 40%, leaving most of the wage gap unexplained. The unexplained gap captures unobserved factors which may be related to productivity and discriminatory practices in the labour market. Women in lower wage jobs face different challenges than women in the top paid jobs. There is strong evidence of “sticky floors” in Indonesia in the formal sector – that is women at the lower end of the wage distribution facing a much bigger gender wage gap than women in higher wage jobs. There is some evidence of this improving over time in the formal sector, as the magnitude of the wage gap at the lower end of the wage distribution is smaller for younger women. In the informal sector the gender wage gap is relatively constant along the wage distribution, although sticky floors appear to be starting to emerge for younger women. Women challenges in the labour market are still important in the Indonesian context besides the existing labour laws that have specifically stipulated antidiscrimination principles. Industrial segregation by gender is a major cause of the gender wage gap in Indonesia – both in the formal and informal sector. This suggests the need for further research to explore the factors that play a role in women’s industry selection and ways to either encourage women to participate in male dominated industries or to improve labour skills of women to be more productive in their current industries. In the formal sector – suggesting the need for research on how minimum wages affect gender wage gaps, particularly for younger workers. Marriage and child-rearing are associated with greater wage gaps, particularly in the informal sector and at the lower end of the wage distribution in the formal sector. That this is decreasing in the younger cohorts is promising but policies that increase women’s ability to easily move between the home and the workplace – for example, provision of child care and public transport and transport infrastructure that takes into account women’s needs – could work to reduce the wage penalty

4

associated with married life and women’s reproductive role. Examining the role of these ways to facilitate women’s ability to juggle work and home life is likely to be a fruitful area for future research. Future Priorities We conclude with a diagrammatic synthesis of our findings which relates potential contributing factors to different facets of gender inequality and ranks them in terms of severity. A number of areas are identified for action. The role of caring responsibilities (and child care) is an area worthy of further attention. Improving our understanding of gendered barriers to entrepreneurship and business expansion is also important and worthy of examination. Similarly, closely examining entry and re-entry decisions into the labour market and their interaction with caring responsibilities and movements in the sector of employment would generate new understandings of women’s participation. Finally, the role of infrastructure, particularly transport infrastructure, in providing women with access to jobs in different occupations/industries is poorly understood and rarely studied. These are all areas we propose for future research. Indonesia has 86.3 million women of working age who do not work outside the home. This is considered as a great waste of resources that could be a considerable engine for economic growth. Gainful employment can also widen women’s horizons, build confidence and ensure that women’s voices are heard outside the home. Understanding the constraints that women face in the labour market and ways in which women can be encouraged to participate and supported in their endeavours is thus of primary importance.

5

List of Contents 1.

Introduction .................................................................................................................................. 10

2.

Overview of Gender Inequality ..................................................................................................... 11 2.1

Educational Inequality .......................................................................................................... 11

2.1.1 2.2

3.

Labour Market ...................................................................................................................... 15

2.2.1

Labour Force Participation, Employment and Unemployment .................................... 16

2.2.2

Employment Status (Formal/Informal) ......................................................................... 19

2.2.3

Industrial and Occupational Segregation ...................................................................... 20

2.2.4

Working conditions ....................................................................................................... 21

2.2.5

Wages............................................................................................................................ 22

2.2.6

Migration....................................................................................................................... 26

2.3

Finance & Entrepreneurship ................................................................................................. 26

2.4

Infrastructure ........................................................................................................................ 29

2.5

Health .................................................................................................................................... 30

2.6

Institutions & Laws................................................................................................................ 32

2.6.1

Law in relation to families ............................................................................................. 32

2.6.2

Labour Laws .................................................................................................................. 33

2.6.3

Property Rights.............................................................................................................. 33

2.6.4

Political Representation ................................................................................................ 34

Stagnation of the female labour force participation in Indonesia: An age and cohort analysis .. 35 3.1

Introduction .......................................................................................................................... 35

3.2

Data and Methods ................................................................................................................ 35

3.2.1

4.

Education attendance and completion ......................................................................... 11

Descriptive results ......................................................................................................... 37

3.3

General results ...................................................................................................................... 38

3.4

Age and cohort results .......................................................................................................... 40

3.5

Conclusions ........................................................................................................................... 41

Gender Wage Gap in Indonesia - a distributional analysis of the formal and informal sector .... 42 4.1

Introduction .......................................................................................................................... 42

4.2

Data and Method .................................................................................................................. 42

4.3

Results ................................................................................................................................... 48

4.3.1

Decomposition across the Wage Distribution .............................................................. 49

4.3.2

Age Cohort Analysis ...................................................................................................... 51

4.4

Conclusions ........................................................................................................................... 53

6

5.

Conclusions and Future Research Agenda .................................................................................... 54

References ............................................................................................................................................ 57 Appendix 1: Blinder-Oaxaca Methodology ........................................................................................... 60 Appendix 2: Gender wage gap along the distribution by status of employment ................................. 61 Appendix 3: Gender wage gap along the distribution by status of employment and age cohorts ...... 66 Endnotes ............................................................................................................................................... 69

7

List of Figures, Tables and Equations Figure 1 Level of school completion by age cohort and gender, 2013 ................................................. 12 Figure 2 Years of education by age cohort and gender, 2013 .............................................................. 12 Figure 3 Literacy rates by region, gender and age cohort, 2013 .......................................................... 13 Figure 4 Indonesia PISA 2012 test results by gender ............................................................................ 14 Figure 5 Total Employment in Agriculture ............................................................................................ 15 Figure 6 Female Labour Force Participation by Country ...................................................................... 16 Figure 7 Labour force participation by gender and age group in 2013 ................................................ 17 Figure 8 Female Unemployment .......................................................................................................... 17 Figure 9 Underemployment by gender and Urban/Rural..................................................................... 18 Figure 10 Informal Status of Employment by Gender .......................................................................... 20 Figure 11 Informal Status of Employment by Region in 2013 .............................................................. 20 Figure 12 Employment by Industry ....................................................................................................... 21 Figure 13 Workers Wage Female/Male Ratio ....................................................................................... 23 Figure 14 Blinder-Oaxaca Decomposition............................................................................................. 24 Figure 15 Blinder-Oaxaca Decomposition by Sector of Employment ................................................... 25 Figure 16 Female’s age at their first marriage, 2013 ............................................................................ 33 Figure 17 Proportion of seats held by women in national parliaments (%) ......................................... 34 Figure 18 Age and cohort effects .......................................................................................................... 40 Figure 19 Logarithm of the Hourly wages of male and female workers............................................... 43 Figure 20 Histogram of the years of experience and education attainment by gender ...................... 43 Figure 21 Gender wage gap decomposition at the mean by sector of employment ........................... 48 Figure 22 Gender wage gap across the wage distribution by status of employment .......................... 50 Figure 23 Decomposition of the explained component of the gender wage gap across the wage distribution by status of employment .................................................................................................. 50 Figure 24 Gender wage gap across the wage distribution in the formal sector by age cohort............ 51 Figure 25 Decomposition of the explained component of the gender wage gap across the wage distribution in the formal sector by age cohort .................................................................................... 52 Figure 26 Gender wage gap across the wage distribution in the informal sector by age cohort......... 52 Figure 27 Decomposition of the explained component of the gender wage gap across the wage distribution in the informal sector by age cohort ................................................................................. 53

Table 1 Enrolment status by gender for individuals aged 5 to 18 years............................................... 11 Table 2 Type of Employees by Gender of Owner ................................................................................. 27 Table 3 Borrower’s characteristics, by gender ..................................................................................... 28 Table 4 Source of Non-Own Capital and Amount Borrowed from the Bank ........................................ 28 Table 5 Delivery attendance ................................................................................................................. 31 Table 6 Average Number of Children by age cohort ............................................................................ 31 Table 7 Summary statistics of labour force participation and explanatory variables .......................... 37 Table 8 Marginal effects of pooled sample........................................................................................... 39 Table 9 Summary statistics of productivity characteristics .................................................................. 45 Table 10 OLS estimates of Wage by gender and sector of employment.............................................. 46

8

Table 11 Characteristics contribution to the total wage gap at the mean by sector of employment . 49 Table 12 An Analysis of Factors Determining Labour Market Gender Inequality in Indonesia ............ 56

Equation 1 Labour force participation .................................................................................................. 36 Equation 2 Wage equation ................................................................................................................... 44 Equation 3 Blinder-Oaxaca Decomposition .......................................................................................... 60

9

1. Introduction This report presents a general overview of the state of gender inequality in Indonesia. The Global Gender Gap Report 3 (2014) prepared by the World Economic Forum identifies inequality in economic participation and opportunity for women as the most significant gender inequality challenge for the country. Economic participation will thus be the main focus of this study. The report contains three main parts. First, we present a general review of different aspects of gender inequality. We examine the different facets of gender inequality in the following order: i. ii.

iii. iv. v. vi.

Educational inequality Labour Market Inequality a. Labour force participation b. Employment status (formal/informal) c. Industrial and occupational segregation d. Working conditions e. Gender wage gaps f. Migration Entrepreneurship and Finance Infrastructure Health inequality Institutions and Laws

As a result of this first stage of the project and the identification of the most important challenges, the second section presents two pieces of analytical work. The first focuses on the main drivers of female labour force participation (FLFP), exploring the factors that have contributed to FLFP remaining unchanged over the last two decades. The second examines the drivers of the gender wage gap and examines how these drivers differ across the distribution of wages in the formal and informal sectors. We conclude with a section identifying the most important inhibitors of gender equality and suggest areas for future research. 4

10

2. Overview of Gender Inequality 2.1

Educational Inequality

Education is recognised as key to reducing poverty in developing countries and is a significant factor in determining wage gaps between men and women. While in the past there were various reasons for lower levels of female enrolment in education in Indonesia, in particular, distance from schools and early marriage (UN, 2003), gender equality in education in Indonesia has increased markedly over recent years to approach parity (ADB 2006). 2.1.1 Education attendance and completion Women’s educational achievement in Indonesia has made significant progress toward equality with men at all levels of education (Buchori & Cameron, 2007; UNICEF, 2010). The gap between enrolment and attainment between men and women has narrowed to the point of disappearing and there does not appear to be a significant ‘son preference’ for education in Indonesia (Kevane & Levine, 2000), although there is some evidence that in hard time families will cut expenditure on girls’ education before cutting educational expenditure on boys (L. A. Cameron & Worswick, 2001). Table 1 presents figures from the 2013 Indonesia’s National Socio-Economic Survey (Susenas) showing that there is very little difference between school attendance for girls and boys in both urban and rural areas. Girls’ attendance is slightly higher than boys’. Table 1 Enrolment status by gender for individuals aged 5 to 18 years

In school Not currently attending school Never attended schooling Total

Male 80% 9% 11% 100%

Urban Female 81% 8% 11% 100%

Total 80% 8% 11% 100%

Male 77% 12% 11% 100%

Rural Female 78% 11% 11% 100%

Total 77% 11% 11% 100%

Source: Authors calculations. Susenas 2013. This is a relatively recent phenomenon so while for younger women there is very little gender differential, older women have lower education levels than their male counterparts. Using the Indonesian Family Life Survey data and logistic regression analysis, Zhao (2006) found that women in older cohorts were significantly less likely to have attended primary school, but this was not seen in younger cohorts (born after 1973). The larger gender gap in education amongst older cohorts can be seen in Figure 1 below which presents data from the 2013 Susenas.

11

Figure 1 Level of school completion by age cohort and gender, 2013

Source: Susenas 2013. The larger gender education gaps in older cohorts can clearly be seen in Figure 2 which presents average years of education by gender for urban and rural areas separately. In both urban are rural areas educational parity has been attained for cohorts aged 29 and below. Figure 2 Years of education by age cohort and gender, 2013

Source: Susenas 2013.

12

The attainment of gender equality in education is a nationwide achievement. This is true even in the outer regions off Java –Bali as can be seen in Figure 3. Figure 3 Literacy rates by region, gender and age cohort, 2013

13

A number of studies report gender gaps in literacy rates. Haidi (2004) finds that the rate of illiteracy was twice as high for women than for men: 6.26% compared to 13.85%. Azzizah (2014) also finds a gap between female and male literacy which varies by region. In their examination of formal employment and literacy, Gallaway and Bernasek (2004) conclude that women are underrepresented in occupations that are correlated with literacy. It seems that the literacy gap is however also constrained to the older cohorts. Figure 3 above shows that in all regions, there is equality across the genders with regard to literacy in the younger cohorts. Azzizah (2014) focuses on the population who have never attended school. He finds that women are more likely to have never attended school than men and that this gap is bigger in rural areas than in urban areas. Indonesia’s patrilineal system and the emphasis on women’s family responsibilities is evident in the reasons given for not attending school, in particular an emphasis on getting married and a requirement to take care of the family (Azzizah, 2014). Rammohan and Robertson (2012) using the Indonesian Family Life Survey finds female educational outcomes are significantly worse for females from provinces with patrilocal norms (as opposed to matrilocal or neolocal norms). These findings may again reflect persisting gaps in the older cohorts. Table 1 above using the nationally representative Susenas data finds no gender differences in having never attended school among those under 25 years of age. Another important aspect of education equality is equality in the quality of education received. One of the main challenges Indonesia faces in terms of education is its low quality. Looking at the result of PISA test scores in 2012, Indonesia was ranked 60 out of 61 countries in mathematics. Compared to other countries of the region, Indonesia underperforms. Indonesian children aged 15 years have an average score of 375 compared to average scores of 573 in Singapore; 511 in Vietnam; 427 in Thailand; and 421 in Malaysia 5. In science and reading Indonesian scores are very low as well with an average score of 396 and 382, respectively. Figure 4 Indonesia PISA 2012 test results by gender

14

Figure 4 shows the PISA performance of Indonesian students in maths, science and reading by gender in 2012. Level 6 is the level attained by top performers. Level 1 signifies a relatively poor performance. Around 75% of Indonesian boys and girls perform at level 1 or lower, representing a very low achievement for both genders. In science the country’s performance is slightly better, again with no gender gap. Reading skills is where there is the biggest proportion of children in level 3 (close to an average performance), with girls outperforming boys. This gender difference is widely observed around the world with women tending to perform better than men in reading tests.

2.2

Labour Market

The Indonesian economy has been growing steadily over the last few decades (with the notable exception of the period following the 1997 financial crisis). Economic growth has been reflected in significant changes in the Indonesian labour force. The labour force is now significantly more urbanised, less agricultural and better educated than it was three decades ago. For example, While in 1970 26% of the labour force was in urban areas and 74% in rural areas by 2007 the composition was 41% urban and 59% rural (Chowdhury, Islam, & Tadjoeddin, 2009). Figure 5 Total Employment in Agriculture

Labour force participation in Indonesia has increased at a faster rate than the working age population. The age composition of workers has changed – with younger workers now constituting a smaller share, possibly because they are studying for longer. These changes and evolving societal norms have affected the experiences of working women – their ability to find work, the type of work they do and the wages they receive. In this section we first examine female labour force participation over time and its relationship with employment, unemployment and underemployment. Next, we look at the formality and informality of employment including some comparisons by industry and regions. Then we look at the gender gap by industrial sector and occupation. We also look at working conditions as these have changed with Indonesian growth and urbanization. Then we look at gender inequalities in wages, separating rural and urban areas and examining changes over time. We also present the findings in the literature from Blinder-Oaxaca decompositions of wages that seek to estimate the extent to which the gender wage gap is explained by the different characteristics of men and women and to what extent it is due to differential treatment of the genders i.e. discrimination. Finally we look at international migration. 15

2.2.1 Labour Force Participation, Employment and Unemployment The 2014 World Development Indicators show 51.4% of Indonesian women aged 15 and above participating in the labour force (either working or looking for work). This is low by international standards. Figure 6 presents female labour force participation rates for countries in the region (from Cambodia with the lowest GDP/capita to Malaysia with the highest), and Australia, the UK and the US for comparison. Vietnam, similarly a lower-middle income country, has a corresponding rate of 73.0%. Thailand, classified as a middle income country, has a female labour force participation rate of 64.3%. The participation of Indonesian women in the labour market is clearly low in relation to similar countries and its level of development. Figure 6 Female Labour Force Participation by Country

Source: World Bank, 2013.

Further, female labour force participation has remained relatively stable over the past two decades. It increased only very slightly from 50.2% in 1990 to 51.4% in 2013. Male participation increased at a higher rate over this period, from 81.1% to 84.2% (Chowdhury et al., 2009). Female participation is less than two-thirds of the male equivalent. Married women and women with more dependent children have the lowest participation rates (Comola & de Mello, 2012). Not surprisingly, women’s labour force participation declines during their most fertile years. Van Klaveren, Tijdens, Hughie-Williams, and Martin (2010) show that while male labour market participation is highest in the age range of 35-49 years, for females it is highest in the post-child-rearing years (ages 45-59). This is consistent with calculations using data from Susenas from 2013 as shown in figure 7.

16

Figure 7 Labour force participation by gender and age group in 2013

Cepeda (2013), in an analysis for the World Bank, uses information from the Indonesian National Labour Force Survey (Sakernas) 2009 to show that young single women aged 15 to 24 have the highest rate of participation compared to other marital categories in this age range. The aggregate drop in participation on marriage in this age range is an enormous 37.7 percentage points. Interestingly the biggest drop is among married women without children, and after the first child the reduction decreases per each additional child. One of the suggested explanations for this is an anticipatory effect. As women get married they expect to have children immediately so they stop working even before pregnancy. 6 From age 25 to 64 divorced and widowed women with children are the ones with the highest labour force participation. Figure 8 Female Unemployment

Women’s labour force participation decisions however reflect a combination of marital and socioeconomic status, Alisjahbana and Manning (2006). 7 Poorer married women are more likely to

17

participate than married women in non-poor households. Taniguchi and Tuwo (2014) find that higher educational attainment is positively associated with the decision to participate in the labour market. Although female employment in Indonesia has also been slightly increasing since 1990, Chowdhury et al. (2009) show that the share of female employment in total employment has decreased from 38.7% to 35.1% between 1990 and 2006, compared to male employment which increased from 61.3% to 64.9%.8 This is a result of greater increases in males’ labour force participation relative to females’, and female unemployment having increased over this period more than male unemployment. In 2006 the unemployment rate was 13.4% for females and 8.5% for males. This has however improved since with 6.7% of women and 5.7% of men being unemployed in 2012.9 Van Klaveren et al. (2010) show that unemployment affects mostly young and highly educated females, as presented in Figure 8. Further, Alisjahbana and Manning (2006) find that better-off women are more likely to be unemployed with poorer women being more likely to be underemployed (working but wanting to work more). This reflects that better off women can afford to stay unemployed for longer periods while poorer women will take whatever work they can find, often in the agricultural and/or informal sectors. Figure 9 shows that underemployment in 2013 is higher for women than for men in all geographic regions. 10 41% of employed women are underemployed compared to 25% of men (this could include voluntary underemployment) and almost 57% of women in rural non Java-Bali provinces are underemployed compared to 37% of men. These differences in the number of hours worked by women will consequently affect average monthly wage income. It is calculated that underemployment results in gender differences in monthly wage income for formal workers of 28.5% and for informal workers of 50.5% (Cepeda, 2013). Figure 9 Underemployment by gender and Urban/Rural

Greater underemployment amongst women is also found to be associated with age and urban areas. Around 20% of young, female workers in poor urban households worked less than 35 hours and were searching for more work. Taniguchi and Tuwo (2014) examine the relationship between 18

underemployment, marital status and education. They find that marriage decreases the probability of underemployment. Higher education attainment increases the probability of full employment and moderate (as opposed to severe) underemployment. The strongest association is with increases in underemployment. 2.2.2 Employment Status (Formal/Informal) One reason unemployment rates are relatively low in Indonesia is that unemployment is “unaffordable” to poor households and the informal sector expands to accommodate those who cannot find formal sector jobs. Informal jobs are low average productivity and low quality (low pay, no social security, low stability and sometimes unsafe conditions). Economic growth has resulted in growth in formal sector jobs. The formal sector was estimated to have been growing at a rate of 5.8% prior to the 1997 financial crisis. It has since been growing at a slower but not insubstantial rate. Chowdhury et al (2009) estimate the formal sector to have grown at 2.2% since the crisis through to 2008 and the rate of growth has increased further since then. 11 We calculate that in 2013 the informal sector is however still estimated to constitute 75% of total employment. 12 The gender difference in informality of employment has shrunk over time. In 1990 the percentage of working women who were employed in the informal sector was 10 percentage points higher than for men. This gender difference had decreased to 7 percentage points by 2006. 13 Hence, in spite of the increase in education levels amongst women, women continue to be concentrated in informal jobs. This difference is driven mainly by the proportion of female unpaid and casual workers, which is 3 to 1 compared to males. Marriage and dependent children increase the probability of being an unpaid female worker (relative to being a paid worker in the informal sector) and higher educational attainment decreases the probability, Comola and de Mello (2012). Similarly Priebe, Howell, and Sari (2014) show that poverty is associated with the sector of employment. They find that 80% of the women in the poorest households work in the informal sector compared to 34% of the wealthiest women. Within both the poorest and wealthiest categories men’s participation in the informal sector is about 5 percentage points less than women’s. As shown in figure 10 agriculture and fishing is the sector with the highest informality for both males and females. In 2013 the agricultural/fisheries sector accounted for about 34.9% of total employment and 32.8% of total female informal employment. If we restrict our attention to paid workers in the informal sector, women are most likely to be working in housekeeping, as homeworkers and in small microenterprises, where wages, working conditions and job conditions are typically poor (Van Klaveren et al., 2010).

19

Figure 10 Informal Status of Employment by Gender

As different provinces have different levels of development and different employment markets, we also present informality by province in Figure 11. Bali and NTB-NTT have the lowest differences in informality by gender across all regions. Gender differences in the extent of informality are larger in urban areas. Figure 11 Informal Status of Employment by Region in 2013

2.2.3 Industrial and Occupational Segregation Indonesia’s economic boom in the 1980s and 1990s led to a decrease in work force participation in the agriculture sector from 66% in 1971 to 41% in 1997 (Sugiyarto, Oey-Gardiner, & Triaswati, 2006). However, women were under-represented in the shift from agriculture to manufacturing. This was 20

mainly due to the level of education and the type of skills that were required for those jobs. Low educational attainment often excluded women from accessing jobs in manufacturing. However, as the gender gap has been shrinking (as discussed above) and migration from rural to urban areas has continued, the total participation of women and men in the agriculture sector has decreased, while participation of both men and women has increased in the manufacturing sector and the relative participation by gender has become more similar. See Figure 12.14 By 2007, 58.1% and 58.9% of working women and men, respectively, were working in non-agricultural sectors. The service and trade and retail sectors have also become larger during this time with women increasing their share of employment in these sectors. Cepeda (2013) shows that administrative and managerial; and clerical and related occupations are mostly dominated by men. Women’s participation in those occupations is 18.0% and 40.4%, respectively. Traditionally those occupations are associated with higher wages, which implies that the share of the wages in the hands of women is very low. Other occupations where male participation is almost double or higher are production and transport and agriculture and related activities. Services, professional, technical and sales are occupations where both genders participate. As mentioned below, services and sales are occupations that usually are on the lower range of wages and require long hours in the job. Figure 12 Employment by Industry

2.2.4 Working conditions Information on working conditions in Indonesia is limited. Van Klaveren et al. (2010) examine the number of hours per weeks worked by gender, disaggregated by status of employment and industry. Even though the average number of weekly working hours is similar for males and females, 42.8 and 38.2, a larger proportion of men work excessive hours (defined as more than 48 hours per week) 31.8% for males compared to 24.5% for females in 2009. 15 Economic growth has coincided with shorter working days and reduced the differential between men’s and women’s working hours. In 2003 around 50% of all males worked longer than 48 hours compared to 41% of females. Average hours of work for females are less than males across all industries. The longest hours being worked for men are in the self-employed sector. The sectors with the longest working hours for women are housekeepers; wholesale and retail trade; and hotels and restaurants. 21

Unlike the national average, in these particular female-dominated industries, the number of hours worked has increased from 2000 to 2008. Working hours are just one facet, and possibly not the most important, of working conditions. Pinagara and Bleijenbergh (2010) argue that women face a disadvantage in negotiation in Indonesia. This reflects gender roles that are driven to some extent by religious views and patriarchal norms. Perceptions of gender roles affect hiring rates, the potential support women can access and the bargaining power they have to advocate for better conditions at work. Women generally have poorer access to workers unions, fair work agreements and contracts. For example, although Foreign Direct Investment (FDI) has increased women’s job opportunities particularly in manufacturing, FDI in export-oriented industries provides incentives for employers to offer more precarious conditions of employment in order to reduce fixed costs and increase international competitiveness (Siegmann, 2007). Lack of support structures like child care reduces the opportunities for participating in the labour market. Women face cultural, social, economic and religious barriers to employment and fair conditions in employment. This may influence the way younger generations of women perceive their labour market prospects and affect their educational, occupational and employment choices. This reproduces and prolongs the segregation of jobs by gender and results in women being overrepresented in low level jobs with minimal decision-making and few visible safety measures that make women more vulnerable than men (AusAid, 2012; Blackwood, 2008; Elliott, 1994). 2.2.5 Wages The majority of studies that examine gender inequality in the labour market focus on wage inequality. There is a large and growing economic literature looking at the causes of the wage gender gap in different countries. The indicators of gender inequality discussed above - labour force participation, employment and unemployment gender ratios, sector and status of employment – also feed into the wage gap. The gender wage gap ultimately reflects differences between men and women in education, training and skills, experience (reflecting reproductive choices), occupational choice, employment status, labour market choices based on social expectations, and discriminatory hiring and other practices. Methods for estimating the contribution of these different factors and studies that do so will be discussed below. Figure 13 presents the female/male ratio of average hourly wages. The average hourly wage of females in the formal sector is somewhere between 70% and 80% of that of males. 16 This looks to have been improving over time. The figures for the informal sector however show a worsening situation. Because of data limitations in the informal sector, most studies that attempt to explain the raw wage gap have focused on the formal sector.

22

Figure 13 Workers Wage Female/Male Ratio

Gender wage gaps persist across education levels but are smaller amongst the better educated. Females who did not graduate from primary school earn only half of that earned by similar males and a female who has graduated from senior secondary earns on average 79%. Feridhanusetyawan et al., (2001) find that the gender wage gap has an inverted U shape with age reaching the maximum by ages 40 to 50. This is likely to be due to cumulative differences in the amount of work experience achieved by females compared with males – as a result of periods of female non-participation in employment due to child birth and child care. Women earn less than men across all occupations and sectors, and this is true at all levels of education, Taniguchi and Tuwo (2014). The wage gap does vary however by industry with the biggest differences being found in agriculture and services in private households where wages are the lowest and women earn around 64% of the male average wage. In the highest paying sector - finance – where wages are almost double the average wage; women earn 6.2% less than men. In the most female-dominated sectors - wholesale-retail and hotels-restaurants - even though there is a relatively high average level of education, the hourly wage rates are among the lowest due to the longer working days (ADB, 2006; Van Klaveren et al., 2010). Consistent with these findings, Alisjahbana and Manning (2006) find that the average monthly earning of employed females to males (aged 25-59 years) is lowest for the poorest households (one half compared to an average of two thirds across all socio-economic groups). The Blinder-Oaxaca decomposition is the most widely used methodology for determining how much of the gender wage gap is due to differences in observable characteristics between men and women (for example, educational attainment, years of experience, occupation) and how much seems to reflect the mere fact that the worker is female, not male, and which is normally designated as discrimination. The Blinder-Oaxaca methodology is explained in more detail in the appendix 1. Although very widely used across the world, as far as we are aware, there are only a few studies that attempt to determine the portion of the gender wage gap due to observable characteristics and the portion left unexplained in the Indonesian context. 17

23

Some of the explored explanations for those differences are differences in education, experience, age, head of household characteristics, industry, poverty level, rural/urban, employment status (formal/informal) and the effect of foreign direct investment and urbanization (Alisjahbana & Manning, 2006; Cepeda, 2013; Feridhanusetyawan, Aswicahyono, & Perdana, 2001; Pirmana, 2006; Siegmann, 2003, 2007; Taniguchi & Tuwo, 2014). Figure 14 Blinder-Oaxaca Decomposition

The results of these studies suggest that the raw wage gap is still high but has decreased over time. The proportion of the gender wage gap that is unexplained (the discrimination component) has however increased over time as presented in Figure 14. 18 This finding emerges from the comparisons of these studies and appears robust even though the studies use different measures of wages, different specifications and different sources of information. These differences explain some of the variability in the summary presented below. Feridhanusetyawan et al. (2001) 19 is the first study of which we are aware that decomposes the gender wage gap. They use the 1986 and 1997 Sakernas. and estimate that the raw wage gap in the formal sector to be about 0.45 in 1986 and 0.35 in 1997. They present estimates for urban and rural areas separately. In 1986 the raw gap was higher in urban areas (0.53) than in rural areas (0.39). The proportion of the gap attrubuted to discrimination was however lower in urban areas (46%) versus 56% in rural areas. By 1997 they estimate that the raw gap had decreased to 0.35 and 0.33 for urban and rural areas respectively and that the unexplained proportion represented a smaller portion - 30% and 42%, respectively. Pirmana (2006) pools data from the 1996, 1999, 2002 and 2004 Sakernas and calculates a raw difference between male and female real monthly wages of 40%. Forty-two percent of this difference (16.8 percentage points) is found to be explained by differences in endowments (education level, experience, socio-demographic characteristics, economic activity and sector and local and regional characteristics) and 58% (23.2 percentage points) is unexplained or due to discrimination. That means that a woman with similar characteristics to a man will on average be paid 23% less. 20 Taniguchi and Tuwo (2014) 21 use the 2010 Sakernas data. They report a raw wage gap of 30.8% for workers with full employment status. They examine the role of age, hours worked, educational attainment, work occupation, industry and geographical location. They find that the vast majority of

24

the wage gap 93.2% (28.7 percentage points) is due to discrimination with only 6.8% (2.1 percentage points) due to differences in characteristics. Although the gender wage gap is higher in urban areas, the discrimination component is larger in rural areas. 22 A limitation of the studies above is that they only analyse women with formal employment status. Women certainly face wage discrimination in the formal sector but most women work in the informal sector. Cepeda (2013) 23 is the only study that examines the determinants of the wage gap in both the formal and informal sector. As shown in Figure 15 the informal sector is found to have not only a higher raw wage gap but also a larger discrimination effect. The gender gap has decreased over time in both the formal and informal sectors but the unexplained proportion of the gap has increased. In the formal sector the unexplained proportion has risen from 33% of the total gap of 34.9% in 2001 (11.6 percentage points) to 45% (9.86 percentage points) in 2010 and in the informal sector has increased from 75% (35.1 percentage points) to 84% (30.5 percentage points). When looking at the drivers of those differences, she shows that differences in educational attainment below tertiary education explain a substantial amount of the wage gaps in the formal and informal sector. Additionally, the author provides evidence of “sticky floors” being a factor in the setting of women’s wages. For both the formal and informal sector, the biggest gender wage gap was found in the lowest two deciles of the wage distribution and then it decreases over the rest of the distribution. Figure 15 Blinder-Oaxaca Decomposition by Sector of Employment

Additional limitations of the studies discussed above are 1) that the use of the Sakernas survey in many of the studies limits the ability to examine the effect of having children on labour market outcomes as it does not contain information on fertility. It is thus not possible to accurately control for career interruptions due to child-raising. This will lead to an over-estimate of the gender-wage gap due to discrimination 24; 2) even those wage differences due to observables e.g. educational attainment, may reflect discrimination. For example women may choose to invest less in education because they anticipate they will be paid less in the labour market. Similarly, observable differences in experience and education can reflect women’s reactions to cultural norms which result in a shorter and more discontinuous working life. Occupational choice may similarly reflect these socio-cultural factors. If the control variables reflect discrimination, our estimate of the discrimination component will underestimate the true extent of discrimination.

25

2.2.6 Migration Migration to foreign countries for work is an important source of income for women in Indonesia and migration rates have increased over the last few decades, both legal and illegal. 25 The most popular destinations are Malaysia and the Middle East. Other Asian countries, for example, Hong Kong, Taiwan and Singapore are also becoming popular destinations. Most female migrant workers work in the informal sector as domestic helpers (World-Bank, 2010a). In 2011 women made up to 75% of the Indonesian foreign workers (World Bank, 2014). This was prior to the Indonesia’s moratorium on domestic work in Saudi Arabia which was imposed in 2011 and is ongoing. Women’s share of total foreign workers has fallen since but they still outnumber men. In 2014 about 54% of total Indonesian overseas migrant workers were female. It is expected that the proportion of women in total migration will decrease further after the government recently announced (May 2015) to extend the ban on domestic workers to twenty-one Middle Eastern countries from August of this year. Most of the migrant women come from poorer, rural regions of Indonesia. Women and men seek to migrate abroad with the expectation of earning wages that are not attainable in poor rural areas (AusAid, 2012). Rural women account for 44% of total Indonesian international migration, although their share has dropped as a result of the moratoria (World-Bank, 2014). Poverty, unemployment, underemployment and lack of formal education (particularly true for older and poorer women) are the main driving forces behind this high rate of migration. Protection for female migrant workers (in Indonesia prior to departure and on return, and in the destination country) is limited. Consequently, mistreatment and even serious physical abuse by employers is not uncommon. 26 There is a real need to formalise, protect and regulate overseas employment (Silvey, 2004). There have been some efforts to improve the situation but Indonesia’s system for labour migration still works poorly and channels of coordination between the government, the recipient governments and migration agencies are still to be improved (Bazzi & Bintoro, 2015). Most of the workers lack access to legal contracts, financial markets and financial literacy, training and in country support. 27 While the moratoria on domestic work in the Middle East was announced as a measure to protect Indonesian women from exploitation while overseas, in practice it will reduce the opportunities for poorer women, particularly in rural areas, to find gainful employment and a path out of poverty. It will restrict many women to the Indonesian labour market which, as seen above, often discourages women from working and treats them inequitably.

2.3

Finance & Entrepreneurship

It is estimated that in Indonesia only 23% of the Small MicroEnterprises (SMEs) are owned by women (Foundation, 2013). Systematic barriers to entrepreneurship prevent women from economic opportunities worldwide. This can not only limit women’s opportunities for starting businesses but can also confine businesses which are established to remain very small in scale, often operating only in the informal sector.

26

Women’s underrepresentation as entrepreneurs in Indonesia is attributed to various factors. Tambunan (2009) identifies obstacles such as low levels of education and fewer training opportunities for women, household responsibilities (especially for rural women), legal, cultural or religious constraints, and a lack of access to formal credit and financial institutions. A lack of time to complete income generating activities due to caring or unpaid roles can also leave women with fewer opportunities to develop their own livelihoods and can result in vulnerability to insecure or discriminatory situations. Using information from the 2014 Micro and Small Manufacturing Industries Survey (IMK) we find that around 45% of manufacturing business owners are female. The nature of men’s and women’s businesses however, appears to differ dramatically. Women’s businesses are smaller in scale and more informal. Table 2 shows that while 30% of businesses owned by men employ paid (mainly male) workers, only 8% of women’s businesses do. Further, men’s businesses have been formalising at a faster rate than women’s (an increase from 17% of male businesses hiring paid workers in 2009 to 30% in 2014 compared to an increase from 3% to 8% for women over the same period). In contrast, female businesses continue to predominantly be staffed by unpaid female labour. Eighty-four percent of women’s businesses rely on unpaid female workers. These findings are consistent with the observation that many women who do become business owners in Indonesia do so out of necessity as a means supplementing household income when the husband’s income is not enough, Tambunan (2014). Hence, there is often a difference in the aspirations of men and women for their businesses, with self-employed females having a lesser desire to expand and/or formalize their businesses. Table 2 Type of Employees by Gender of Owner

Total Proportion of paid Males proportion of paid Females Proportion of unpaid Males Proportion of unpaid Females

2009 8% 3% 35% 54%

2014 14% 6% 32% 48%

Owner is a male 2009 2014 14% 24% 3% 6% 57% 52% 26% 18%

Owner is a female 2014 2009 1% 2% 6% 2% 9% 7% 84% 88%

Source: IMK 2009 and 2014. Authors’ calculations

For those women who do seek to expand their business, access to, and control of financial assets have been shown to be strongly linked to women’s decision-making power within the household (AusAid, 2012). Unlike in many other developing nations, microcredit in Indonesia has not been specifically targeted towards women (ADB, 2006). A qualitative meta-analysis conducted by Vong et al. (2013) found that inequality in access to microfinance reflects differences in educational attainment and cultural norms rather than characteristics of microfinance itself. While it is often asserted that women are less likely than men to use financial institutions, or formal banks in particular, Dames (2012) similarly finds that it is education rather than gender that is an important determinant of whether Indonesians access credit. 28 The gender gap in education has largely closed but a lack of education could remain a barrier to finance for older women. Many studies link financial participation to education, but also more specifically to financial education. In a later study Vong and Song (2015) cited surveys finding almost half of Indonesian women ‘admitted they are very inexperienced in financial services and their lack of understanding of financial products causes difficulties for them to formulate sound financial decisions’. Access to credit is noted to be more of an issue for rural women and is particularly related to property ownership rights and, consequently, the ability to offer collateral 27

against loans. The cost of bank transactions is also found to explain the gap between female and male financial participation. A ‘one-stop platform’ for transactions to reduce the opportunity cost of time such as childcare, transportation and account identification processes’ is recommended, Vong and Song (2015). There seems to be a disconnect between the findings of small scale studies which document disadvantage for women in accessing finance in Indonesia, and the findings of much larger, more representative surveys which find little evidence on gender differentials, making this an area worthy of further work. AusAid (2012) emphasises the importance of gender considerations when formulating financial inclusion policy, but World-Bank (2010b) notes they have observed few significant differences in gender-disaggregated indicators related to financial inclusion, such as informal savings and having a bank account. World-Bank (2010b), using data from two surveys 29 on access to financial services, also finds few significant gender differences in access to financial services. There were no significant gender differences in borrowers’ characteristics or the institution they choose to borrow from (see Table 3). They did however find gender differences in the reasons given for having a bank account. Women were more likely to have a bank account in order to save for future needs, whereas men were more concerned about their ability to obtain a formal loan. Table 3 Borrower’s characteristics, by gender

Source: World-Bank (2010b). Most female entrepreneurs in Indonesia use personal and family savings as the most common source of capital, Asia Foundation (2013). This is however also true, although slightly less so, of males. The 2014 IMK data show that 88% of women financed their business with their own capital compared to 82% of men. Table 4 shows that for owners who used other sources for capital, 37% of males used a bank loan compared to only 12% of females. Furthermore, if women do borrow for their businesses, the amount borrowed is smaller. Of respondents who did not use a bank loan as a source of capital, 62% of the women reported that the main reason was that they were not interested in borrowing (compared to 45% of men). 30 Table 4 Source of Non-Own Capital and Amount Borrowed from the Bank 2009 Bank Cooperative finacial institution (no bank) Venture Capital Borrowing from Partners Borrowing for people Family Other Total

Male 25% 3% 2% 0% 44% 10% 16% 100%

2014 Female 4% 3% 1% 0% 42% 6% 44% 100%

Male 37% 4% 3% 0% 7% 31% 10% 8% 100%

Female 12% 6% 3% 0% 16% 38% 6% 18% 100%

Source: IMK 2009 and 2014. Authors’ calculations

28

The IMK data do detect a gender gap in access to finance, although maybe not as large as may have been expected, at least in the manufacturing sector. However, it is important to note that the IMK sample provides information only on people who have a manufacturing business and so may not present a complete picture of access to finance in Indonesia. For example, we do not know how many women (and men) wished to start a business but could not do so due to a lack of capital. To the extent that more women than men could not start a business, the IMK data will under-represent the extent of inequality in access to finance.

2.4

Infrastructure

The provision of infrastructure determines the ability of both men and women to produce output and access jobs. This is true of the provision of energy - electricity and gas – which can be necessary to run small businesses, and also transport infrastructure. The gendered impact of the provision of transport infrastructure and services is an understudied area but one which is starting to receive more attention. A number of studies have documented how women’s transport needs differ from those of men. 31 Women have been found to be more dependent on public transport than men as men, as the main breadwinners, are the ones who most often have primary access to any household vehicle e.g. motor cycles, leaving the women in the household to travel on foot or by public transport. Women’s transport needs also differ because they often have responsibility in the household for shopping, caring for children and the elderly, while also possibly working to generate an income. Thus, while men’s transport needs are well met by transport aimed at getting people to and from business centres in peak periods, women’s needs differ. Women require services distributed more evenly throughout the day with routing and ticketing that allows for more stops and stops in areas that allow for the carrying out of household shopping and other chores. Security on public transport is also of greater concern to women. The provision of women’s only carriages goes some way towards reducing the opportunities for sexual harassment on public transport (a frequent occurrence) but other measures, such as providing safe, well-lit and visible waiting areas also make public transport more female-friendly. Like most places, the transport industry is male-dominated in Indonesia, leading to women’s needs being overlooked. There is a growing recognition that greater gender balance is needed amongst transport planners and engineers if we are to see a transport system that balances the needs of men and women. Inadequate transport infrastructure limits the range of employment opportunities people can access and is likely to have a disproportionately large effect on women given their greater reliance on public transport and the need for safe and reliable transport that enables them to fulfil their employment and household responsibilities. Although gender differences in transport needs have started to be recognised (particularly in relation to urban transport), there is very little work on the impact of transport infrastructure and female labour market participation. By providing physical access to jobs and markets transport infrastructure can play a potentially important role in boosting women’s economic participation32.

29

2.5

Health

Gender differences in health status are important in their own right and also affect women’s ability to participate in the labour market and their productivity. A lack of investment in women’s health has long lasting consequences, affecting cognitive development, school progression and labour income. Gender gaps in health early in life are likely to widen from childhood to adulthood. Gender gaps in health – for example in infant and child mortality and morbidity – although found in many developing countries, are not evident in Indonesia. There is some evidence that women have more limited access to curative medical treatment than men (reflecting their lower capacity for payment) (ADB, 2006). There is however not much information on women’s access to general health services. Mental health is one area where women seem to be more in need than men. Using information from the Indonesian Family Life Survey (IFLS) Friedman and Thomas (2009) find that women are more likely than men to report feeling sad and anxious and having difficulty sleeping. They are also more likely to report being in poor health. 13.6% of women 15-49 years suffer from a chronic lack of protein and anaemia (JICA, 2011). Indonesia does not perform particularly well in respect to reproductive health. Maternal mortality rates are high (although decreasing) relative to similar countries. The 2013 maternal mortality rate was 190 per 100,000 live births (falling from 450 in 1986 and 307 in 2000 (ADB, 2006)) while other members of the region like Vietnam register only 47 per 100,000 live births.33 Maternal health services in Indonesia are generally of low quality. Table 5 presents the person who attended the first and the last delivery of women over time. When examining the decisions of the same woman across time, we can see that from the first to the last delivery more women were going to see a doctor or midwives and also fewer women went to traditional healers or family members. This pattern is apparent in both, rural and urban areas. When comparing across years, we can see that the proportion of women whose last delivery was attended by a medical professional increased from 73% in 2007 to 85% in 2013. This increase is much stronger in rural areas where the professional assistance at birth increased 16 percentage points, compared to 4 percentage points in urban areas. Although deliveries assisted by trained personnel have increased, the maternal mortality rate is persistently high in Indonesia. The poor quality of care is the likely culprit. Non-professional assistance still accounts for 23% of care in rural areas. Access to contraception is also very limited for non-married women. Although the fertility rate has been decreasing over time (Table 6), in 2004 the contraceptive prevalence rate was only 60% in 2004 (JICA, 2011) 34 and most of the methods were women biased, for example oral contraceptives and injectables, as opposed to condoms. A further area where little is known, is the health status of elderly women. Life expectancy at birth is 5 years higher for women than men (68.8 for men and 72.7 for women). This can be of particular importance as aging requires specific health care and as female life expectancy is 5 years higher than men’s, women are at greater risk of a lack of provision of services in old age. This is particularly true for women in the poorest households, where 10% of households have a female household head and the average age is 55.

30

Table 5 Delivery attendance

Person who attended the delivery Total Doctor Midwife Paramedic Traditional Healer Family Other Total Urban Doctor Midwife Paramedic Traditional Healer Family Other Total Rural Doctor Midwife Paramedic Traditional Healer Family Other Total Source: Susenas

2007 First Last delivery delivery

2013 First Last delivery delivery

2011 First Last delivery delivery

12.32 53.96 0.52 30.27 2.69 0.24 100

13.64 58 0.89 25.31 1.91 0.25 100

15.44 61.85 0.39 19.79 2.33 0.2 100

16.88 63.71 0.66 17.34 1.24 0.16 99.99

16.61 65.04 0.42 15.46 2.34 0.13 100

18.21 66.02 0.53 13.79 1.37 0.08 100

20.71 64.25 0.39 13.4 1.12 0.13 100

22.25 65.81 0.64 10.51 0.66 0.13 100

23.24 65.82 0.32 9.68 0.83 0.12 100.01

24.87 65.48 0.54 8.74 0.28 0.1 100

24.42 66.89 0.45 7.29 0.84 0.11 100

26.01 66.44 0.39 6.7 0.39 0.07 100

6.11 46.35 0.61 42.76 3.86 0.32 100.01

7.27 52.22 1.07 36.27 2.83 0.34 100

7.9 58.02 0.47 29.55 3.79 0.27 100

9.15 62 0.78 25.66 2.18 0.23 100

9.12 63.27 0.4 23.29 3.78 0.15 100.01

10.73 65.62 0.66 20.58 2.32 0.09 100

Table 6 Average Number of Children by age cohort

15-19 20-29

Age

30-39 40-49 50-59 60-69 70 or more

2007 0.5 (0.6) 1.3 (0.9) 2.5 (1.4) 3.5 (2.0) 4.4 (2.5) 5.1 (3.0) 5.3 (3.2)

Year 2011 0.5 (0.6) 1.2 (0.9) 2.3 (1.3) 3.3 (1.9) 4.1 (2.3) 4.9 (2.8) 5.2 (3.1)

2013 0.5 (0.5) 1.2 (0.8) 2.3 (1.3) 3.2 (1.8) 4.1 (2.3) 4.8 (2.7) 5.3 (3.1)

Source: SUSENAS 2007, 2011 and 2013. Standard Deviation in parenthesis.

31

2.6

Institutions & Laws

Commitments to improving and achieving gender equality can be demonstrated through laws, national and regional policies as well as institutions. Indonesia ratified its commitment to the UN Convention on the Elimination of All Forms of Discrimination against Women (CEDAW) in 1984 and has subsequently reconfirmed its position through its support of subsequent declarations such as the Beijing Declaration (UN, 2003). A number of labour laws in Indonesia deal directly with gender equality. For example, laws governing maternity and menstruation leave. Others target the overall population like the establishment of minimum wages. Many of these laws are not enforced. Those that are enforced, have sometimes proven to have unforeseen negative consequences for women. For example, Suryahadi, Widyanti, Perwira, and Sumarto (2003) find that the imposition of minimum wages between 1998 and 2000 had a negative effect on the employment of low-skilled women from poorer households. Similarly, it has been suggested that the maternity leave provisions enshrined in Indonesian law act as a disincentive for employers to formally hire women. The effect of laws or laws changes is an understudied area in Indonesia that could shed light on how to promote gender equality in the labour market. For example, looking at the effect of minimum wages; menstrual, miscarriage and maternity leave provisions; or the Equal Employment Opportunity strategy implemented in 2003 on FLFP, status of employment and wage gaps. Despite support for international conventions, some laws in Indonesia do not have equal impacts on women and men. Some laws actively limit women’s independence. For example, Indonesian tax regulations require married women to use the same tax file number as their husband (ADB 2006), making it more difficult for married women to make independent financial decisions. Additionally, the Civil Code requires husbands to assist women in signing contracts, removing women’s control over their own financial transactions. 2.6.1 Law in relation to families Documents proving head of the household status are required for female-headed households to access government poverty relief programs and other entitlement programs as well as procuring birth certificates for their children, which are required for state school enrolment. Women in poor families however often lack such legal documents (such as divorce certificates, Alfitri (2012); or identification cards, Lockley, Tobias, and Bah (2013)). Extensive legal reform in Indonesia has taken place to increase women’s access to the religious courts in order to formally document their role as the head of the household (Alfitri, 2012; World-Bank, 2011). Such legal reforms were required as the cost of court fees and transportation to access the courts was, and in some cases remains, beyond the means of the poor (World-Bank, 2011). Although the legal age for marriage in Indonesia is 21 years old, with parental permission women can be married as young as 16 years old, ADB (2006). Early marriage can lead to leaving school before finishing, as many educational establishments will not accept married women, as well as adolescent pregnancy and its associated risks (ADB 2006). Figure 16 presents the distribution of age at first marriage. It shows a significant proportion of girls get married before the age of 18, particularly in rural areas. The average age of marriage has been increasing very slowly from 19.5 in rural areas in 2007 to 19.77 in 2013. 32

Figure 16 Female’s age at their first marriage, 2013

Source: Susenas 2013. 2.6.2 Labour Laws Labour laws are another, unintended source of discrimination against women. ADB (2006) suggests that the high unemployment rate amongst Indonesia women may be the result of labour laws which provide provisions for women suffering from menstrual pain to take leave as well as maternity and miscarriage leave. Maternity leave provisions state that a female worker shall receive her wages in full for the period of maternity leave (Van Klaveren et al., 2010) yet in practice women are dismissed rather than afforded their three months maternity leave (ADB 2006). Elliott (1994) argues that even though maternity leave provisions are strong, due to the nature of the work women are employed in - low paid, unskilled, and seasonal – and the oversupply of labour for these positions, women can easily be dismissed. In the creation of labour laws after independence, Indonesia based many of their laws on colonial legislation and in some cases enforced traditional gender ideologies (Elliott, 1994). In particular, the author cites the exclusion of women from night-time work as one of the prohibitive aspects of the Labour Act (1948). There have been legal attempts to promote equality of remuneration by gender; such as the Equal Employment Opportunity strategy implemented in 2003. However the enforcement of those regulations is not strong enough to be effective (Pinagara & Bleijenbergh, 2010). 2.6.3 Property Rights Access to land and property rights form the productive basis of many households in parts of Indonesia where small-scale agriculture is often the primary food and income source. Women’s access to land or registration of the title in their name is uncommon, with the majority of marital property being registered in the husband’s name (ADB 2006). Although joint land ownership is formally adopted in the law and co-ownership is informally recognised, few land titles are held jointly.

33

Inherited property in Indonesia is allocated predominately according to Islamic law (ADB 2006), which allocates greater proportions to sons rather than daughters; although there are some regions that emphasise gender equality in inheritance as per adat traditions (as shown by responses to the Indonesian Family Life Survey), Kevane and Levine (2000). Furthermore, there are regions, Java in particular, where the youngest daughter fulfils the caregiving role to older parents and as such inherits the parental home. In regions with matrilineal traditions the property is passed from mother to daughter (Kevane & Levine, 2000), however this is not particularly common. 2.6.4 Political Representation Female political representation is often seen as a way to ensure policy decisions are made with gender equality in mind. The Government of Indonesia has set targets for women’s participation in parliament, political parties and decision-making institutions, with legislation mandating 30 per cent female representation (JICA, 2011). However, these levels have not been achieved and are now a target of the current National Mid-Term Development Plan. The Constitution of Indonesia promises equal protection to all citizens, but the Indonesian local governments are male-dominated (Kevane & Levine, 2000). Figure 17 shows that, although increasing, the number of seats is below the government’s target and also low by international standards. Other countries like East Timor, the Philippines and Vietnam have 38.5%, 27.7% and 24.3% female representation in parliament, respectively. Figure 17 Proportion of seats held by women in national parliaments (%)

Women’s voices are also underrepresented in corporate boardrooms (World Policy Analysis Centre (2015); Credit Suisse (2014)). In 2013, only 5% of positions on company boards in Indonesia are held by women. This compares unfavourably with the global average of 12.7% and in comparison to other countries in the region like Malaysia (11%) and the Philippines (12%). Political power was radically decentralised in Indonesia in 2001, with significant decision-making powers being transferred from the national government to district governments. The reintroduction of traditional laws and institutions has been the most cited effect of decentralisation on women (Mahy, 2012). The ambitious transition to a decentralised system of governance has “unintentionally made way for a number of local governments to advance their aspiration of public policies based on Shari’a or Islamic law,” (ADB 2006, p28). Mahy (2012) explains that some of these local laws are particularly discriminatory towards women in particular, not recognising the right of women to own property or earn an independent income. Siahaan (2003) echoes this finding on the effect of decentralisation on women, noting that although it has increased participation ‘it has been less encouraging to women’s participation and [political] representation at the local level’.

34

3. Stagnation of the female labour force participation in Indonesia: An age and cohort analysis35 3.1

Introduction

Indonesia now has the largest economy in the Association of Southeast Asian Nations and the 16th worldwide (ADB, 2015). The continued economic development has meant rising average incomes, changes in the sectoral structure of the economy (from agriculture to manufacturing and services) and increasing industrialization and urbanization (Elias & Noone, 2011). In spite of the significant changes, the impact on the experiences of women in the labour market appears to be rather muted. The 2014 World Development Indicators show 51.4 per cent of Indonesian women aged 15 and above participating in the labour force (either working or looking for work). This has remained largely unchanged over the past two decades which has meant that the large gap between female and male participation continues and female participation remains low relative to countries at a comparable stage of development in the region (see also ADB, ILO, and IDB (2010)). In section 2, we reviewed studies that were largely consistent in identifying the main drivers of female participation. These include marital status, prevailing economic conditions and the level of educational attainment. The main aim of this section of the report is to disentangle how the drivers of female labour force participation in Indonesia and have contributed to keeping female labour force participation unchanged over the period 1996 to 2013. We do this by separating labour force participation into components due to factors on the supply and demand side of the labour market – educational attainment, marital status, fertility, household structure, distance to urban centres, main local industries – and implementing a cohort analysis which separates out the effect of life-cycle factors (age) on women’s labour market participation and cohort effects (changes in participation over time). Understanding the constraints that women face in the labour market is essential in informing policies aimed at addressing these constraints. Previous studies attribute this to gender differences in family roles, child-caring and also cultural norms in relation to women’s traditional roles (Jayachandran, 2014). Increases in participation are likely to have flow on effects through female empowerment and may affect other facets of the gender divide (e.g. political representation, having greater say over household decisions and being less accepting of spousal violence). Improving female participation is also important to help the Indonesian economy shift from a pattern of economic growth driven by resources and cheap labour and capital to growth based on high productivity and innovation (ADB, 2015). This could help Indonesia avoid the middle-income trap and continue its economic development into the future.

3.2

Data and Methods

The data used in this section is from two sources - the National Socioeconomic Survey (SUSENAS) and the Village Potential Statistics (PODES). The SUSENAS is a nationally representative survey conducted annually and typically composed of about 200,000 households. Each survey contains a core questionnaire which consists of information on all household members listing their sex, age, marital status, and educational attainment and information on labour market activity, health and fertility.

35

The Susenas allows us to explore the role of child-raising in the decision to participate and the availability of alternative child-carers in the household (primarily grandparents and other women who could act as babysitters). We supplement the Susenas data with data from the PODES. This is a census of all villages across Indonesia (approximately 65,000). We use the PODES for some demand side characteristics of the labour market such as the distance to the nearest district office (to act as a proxy for access to jobs) and the main source of income of the village. •

•

•

At the individual level, we control for if the individual is the household head, their marital status (e.g. married, divorced, widowed or single) and the level of education achieved by the individual as measured by their receipt of certificate (e.g. if the individual completed primary school, lower secondary school, upper secondary school, or tertiary education). At the household level, we control for the number of people living in the household, the number of females aged between 45 and 65 years in the household (excluding the female respondent) who are potential babysitters, the number of elderly (defined as greater than or equal to 65 years of age) females or males in the household and the number of children in the household by age (the age groupings are 0 to 2 years of age, 3 to 6, 7 to 11, and 12 to 17). At the village level, we control for distance to the nearest district office and the main source of village income. We also control for provincial unemployment rates (calculated from the Susenas) to act as a proxy for the underlying economic conditions at that time.

A disadvantage of the Susenas is that it is cross-sectional so we cannot observe the same individuals or households across time. But by using the Susenas from 1996, 2000, 2007, 2011 and 2013 survey years, we can observe how the participation of different birth cohorts (groups of people born in the same years) change over time. Using data covering such a long time period allows a close examination of lifecycle (age) effects and trends over time (cohort effects) on female participation. To estimate the determinants of female labour force participation we regress whether an individual participates in the labour force or not (yi) on a set of potential drivers (xi) using a binary probit model. That is, we estimate: Equation 1 Labour force participation

yi= β0 + ∑𝑘𝑘𝑖𝑖=1 𝛽𝛽𝑖𝑖 𝑥𝑥 i + εi , y= 1 [y* > 0],

The vector of potential drivers (xi) includes those discussed above. On the supply side of the labour market we control for marital status, if the individual is the household head and the highest level of education achieved, household size, the presence of a babysitter or elderly men or women in the household and the number of children at certain ages. On the demand side, we include distance to the nearest district office and the main source of income in the village. We also control for geographic differences using province dummies and the unemployment rate for each province. Intuitively, the regression identifies the relationship between the control variable and labour force participation. The magnitude of the effect is captured by the coefficient on the control variable (β). Dummy variables are also included for the age of the individual at the time of the survey and their year of birth. The age and cohort analysis will establish whether the younger cohorts behave differently in relation to labour force participation compared to their older counterparts and the extent to which the propensity to participate in the labour market has changed over time. The

36

coefficients (and associated marginal effects) on the age dummies capture how an individual’s likelihood of participating varies across the life-cycle, irrespective of their year of birth after controlling for other characteristics. The coefficients on the year of birth dummy variables allows us to compare people born in different years and so identify whether the younger cohorts behave differently in relation to labour force participation than their older counterparts. 36 We estimate equation (1) separately for men and women and disaggregated by rural and urban and Java-Bali and non-Java-Bali to give us an understanding of the main drivers of female participation. 3.2.1 Descriptive results Table 7 presents the summary statistics of labour force participation d the explanatory variables for rural and urban areas. Table 7 Summary statistics of labour force participation and explanatory variables

Urban Female

Variables Male Individual characteristics: Labour force participation 81.2% 47.3% Household head 57.3% 7.5% Marital status: Single 37.1% 71.2% Marital status: Married 61.1% 63.4% Marital status: Divorced 0.9% 2.6% Marital status: Widowed 00.9% 5.2% Education: At least primary 90.8% 86.0% Education: At least lower secondary 69.5% 62.3% Education: At least upper secondary 22.1% 18.5% Education: At least tertiary 10.5% 9.5% Household characteristics: Household size 4.8 0.3 Babysitter Number of elderly females 0.1 Number of elderly males 0.1 Number of children: 0 to 2 years old 0.2 Number of children: 3 to 6 years old 0.3 Number of children: 7 to 11 years old 0.4 Number of children: 12 to 17 years old 0.7 Village characteristics: Distance to nearest district office ('100km) 0.5 Main income: Agriculture 0.3 Main income: Mining/quarrying 0.01 Main income: Processing/industry 0.1 Main income: Large trading/retail 0.2 Main income: Services other than trade 0.35 Unemployment 0.06 Observations 469,157 481,751 Source: Author’s calculations using Susenas and PODES.

Male

Rural Female

88.5% 62.1% 30.8% 67.0% 1.0% 1.2% 75.4% 38.5% 8.1% 2.8%

56.1% 6.7% 19.9% 71.8% 2.6% 5.7% 67.2% 30.8% 6.2% 2.5% 4.7 0.3 0.1 0.1 0.2 0.4 0.5 0.7

0.8 0.961 0.01 0.01 0.01 0.02 0.06 681,427 691,280

37

As described above, there is a substantial gap between female and male labour force participation – female labour force participation is on average 40 per cent less than male participation (85 per cent compared to 52 per cent). The participation rates are higher for men and women in rural areas compared to urban areas. Most household heads are males, and most females and males are married. There are more potential babysitters in urban households, possibly due to higher housing prices. At the village level, the distance to nearest district office is unsurprisingly less in urban areas and agriculture is most prevalent in rural areas while services and large trading/retail are large income sources in urban areas.

3.3

General results

Table 8 presents the results of estimating equation (1) for men and women by rural and urban status. Marital status is a key driver of labour force participation for women. A married woman in rural areas is 11 percentage points less likely to be working or looking for work than a single woman and this difference is statistically significant. The impact is more pronounced for married women in urban areas as they are 25 percentage points less likely to be participating than single women. Being a household head for both men and women increases the likelihood of participation in both urban and rural areas. But the magnitude of the impact for men is substantially smaller because men are generally the primary income earners so work irrespective of whether they are the household head or not. The level of educational attainment is also a strong driver of female labour force participation. For women, completing upper secondary school increases the likelihood of participation compared to someone who only completed lower secondary by 19 per cent in rural areas and by 22 per cent in urban areas respectively. The magnitude of the impact increases further if women attain tertiary education. But for men there is little variation in the probability of participating with different levels of education. Men, as the main bread winners in Indonesian society tend to work, regardless of their level of education. Household size decreases participation for women in rural areas –an increase in household size of one decrease the likelihood of participation by nearly 2 percentage points. But the magnitude of the impact for urban females and males are much closer to zero. The presence of a potential babysitter, elderly female or male in the household significantly increase the likelihood of female participation by around 1 to 3 percentage points. This may reflect the ability of the woman to leave children at home with the babysitter or the elderly relative. The magnitude of the impact of these potential childminders is much higher for women than men (the effect is negligible for men). The presence of children in the household also has markedly different effects for men and women. For women, the presence of young children has a negative effect on the likelihood of participating. The presence of a child under two years of age decreases the probability of participation by 8 percentage points but has only a small (and positive) effect on men’s labour market activity. On the demand side of the labour market, we hypothesised that the coefficient for distance to the nearest district office would be negative as it was intended to capture distance to an active labour market, however, the coefficient is positive, albeit small. The variable could be positively correlated with agricultural employment in rural areas, with the positive coefficient reflecting women’s greater involvement in agriculture. The villages’ main sources of income variables show that female participation is highest in areas with agriculture and industry (which includes manufacturing). But as the economy moves further away from agriculture to other sectors, female participation drops.

38

Table 8 Marginal effects of pooled sample

Rural Variables Household head Marital status: Single (omitted) Marital status: Married Marital status: Divorced Marital status: Widowed Education: No schooling (omitted) Education: At least primary Education: At least lower secondary Education: At least upper secondary Education: At least tertiary Household size Babysitter Number of elderly females Number of elderly males Number of children: 0 to 2 years old Number of children: 3 to 6 years old Number of children: 7 to 11 years old Number of children: 12 to 17 years old Distance to office ('100km) Main income: Agriculture (omitted) Main income: Mining/quarrying

Urban

Female 0.2115*** (0.0031)

Male 0.0563*** (0.0015)

Female 0.1191*** (0.0040)

Male 0.0385*** (0.0021)

-0.1129*** (0.0025) 0.0057 (0.0050) -0.1629*** (0.0046)

0.0758*** (0.0016) 0.0089*** (0.0019) 0.0146*** (0.0016)

-0.2508*** (0.0028) 0.0059 (0.0058) -0.1626*** (0.0047)

0.1579*** (0.0025) 0.0304*** (0.0033) 0.0491*** (0.0025)

-0.0297*** (0.0016) -0.0652*** (0.0017) 0.1257*** (0.0032) 0.2516*** (0.0040) -0.0158*** (0.0006) 0.0177*** (0.0020) 0.0315*** (0.0025) 0.0252*** (0.0024) -0.0792*** (0.0016) 0.0055*** (0.0013) 0.0251*** (0.0012) 0.0223*** (0.0011) 0.0011 (0.0009)

0.0018** (0.0007) -0.0404*** (0.0007) 0.0169*** (0.0009) 0.0062*** (0.0019) -0.0049*** (0.0002) 0.0049*** (0.0005) 0.0036*** (0.0009) 0.0088*** (0.0009) 0.0105*** (0.0007) 0.0084*** (0.0006) 0.0088*** (0.0004) 0.0073*** (0.0004) -0.0036 (0.0028)

-0.0226*** (0.0026) -0.0608*** (0.0020) 0.1609*** (0.0027) 0.2038*** (0.0034) 0.0046*** (0.0006) 0.0134*** (0.0022) 0.0106*** (0.0029) 0.0209*** (0.0030) -0.0754*** (0.0020) -0.0251*** (0.0016) -0.0043*** (0.0014) 0.0041*** (0.0012) 0.0159*** (0.0016)

0.0175*** (0.0021) -0.0632*** (0.0011) 0.0588*** (0.0012) -0.0085*** (0.0024) -0.0040*** (0.0004) -0.0059*** (0.0011) -0.0039** (0.0017) 0.0060*** (0.0019) 0.0183*** (0.0014) 0.0170*** (0.0011) 0.0152*** (0.0009) 0.0115*** (0.0007) 0.0005 (0.0019)

-0.0695*** (0.0077) 0.0047 (0.0030) -0.0180*** (0.0021) -0.0389*** (0.0020) -0.0070*** (0.0002) 481,751

-0.0172*** (0.0014) -0.0307*** (0.0013) 0.0499*** (0.0018) 0.0756*** (0.0011) -0.0015*** (0.0120) 469,157

-0.1260*** -0.0289*** (0.0103) (0.0034) Main income: Processing/industry -0.0191*** -0.0325*** (0.0071) (0.0025) Main income: Large trading/retail -0.0942*** 0.0244*** (0.0069) (0.0006) Main income: Services other than trade -0.1307*** 0.0366*** (0.0048) (0.0004) Unemployment -0.0027*** 0.007*** (0.0002) (0.0059) Observations 691,280 681,427 Source: Authors calculations using Susenas and PODES. * The marginal effects for province and age dummies can be provided on request. Significance levels *** p