Female labour market participation is the best available indicator for the economic role of women when the purpose is to take into consideration several decades and over 150 countries. Surely it would be interesting to measure the impact of growth on economic empowerment of women and not only on their labour market participation. Examining the real economic role of women would not only entail measuring female labour market participation, but also taking into account women’s working hours (full time, part time), women’s qualification and income, their detailed working conditions (precarity, degree of social security) and their part of decision making in the economy, in politics and society, for example.
However, the limited data availability makes it impossible to include all this information in the estimation model. Firstly, it is not possible to create an “female empowerment index” for all 184 countries and four decades, because available data are fragmentary. Missing observations, especially for development countries and early time periods, would bias the estimation results to a great extent. Secondly, creating an indicator for the empowerment of women based on several variables weakens the overall view and the transparency of the estimation results. Nevertheless, there exist several wider measures of female empowerment today. The United Nations Development Program, for example, offers a Gender Related Development Index (GDI), which represents a gender specific Human Development Index (HDI)17 that tries to capture women’s well being. The GDI takes into account women’s life expectancy, education and income and penalises gender disparities in these fields. The GDI index is available for a large set of countries, but for a large set of countries the database contains observations only from 2006 on, and therefore the data can not be used in the panel data framework. The same problem applies also to another existing index which is the GEM (gender empowerment measure), a composite indicator that captures gender inequality in political participation and decision-making (women’s and men’s percentage shares of parliamentary seats), economic participation and decision-making power (women’s and men’s percentage shares of positions as legislators, senior officials and managers and women’s and men’s percentage shares of professional and technical positions) and power over economic resources (women’s and men’s estimated earned income in PPP US$). Hence, in order to keep the panel data framework it is advisable to concentrate on the female labour market participation as indicator of the economic role of women.
Nevertheless, even simple measures of female labour market participation are confronted with serious measurement problems. Data weakness arises because female work is often informal and therefore unrecorded. In developing countries, the major part of women works informally in the agricultural and in the black market sector (c.f. Chen et al. 1999). In addition, non-paid work, domestic and family-related activities and self-employed work are rarely included in the statistics (c.f. Waring, 1988; Klasen, 2002). The UNDP (1995) shows that 66% of the female activities in developing countries are not captured by national accounts, compared to only 24% of male activities.
It would also be useful to differentiate the female labour market participation by employment status, by sector, by working hours, by age or by the presence and number of children, and to separate female labour market participation into wage workers, unpaid family workers and self-employed workers. But there are large inconsistencies in the definition of specific labour market participation measurements across the world. Taking into account differentiated measures of female labour market participation would imply reducing the observations to developed countries of the Western World.18 Moreover, even data on general female labour market participation has to be handled with care, because the measurement methods and definitions can differ in quality and coverage between countries and data sources. Differences in statistical and conceptual practices lead to a gender bias in official statistics and concepts of labour market participation (c.f. Bardhan and Klasen, 1999). Hence, even when focusing on female labour market participation only, empirical findings can be biased due to measurement problems and the regression results must be interpreted with care.
In view of these measurement problems, I use three alternative empirical specifications for female labour market participation to test the robustness of the findings (sensitivity analysis):
• FLF: The female labour force in percentage of the total labour force. As the total labour force includes both men and women, measurements of the female labour force take into account the level of the male labour force.
• FAR: The female activity rate as share of female formal sector employees in the female working age population (in percentage points). This measurement does not account for the male activity rate.
• RAR: The ratio female activity rate / male activity rate
Table 10 in the appendix gives an overview of the data used in this paper. The following sections describe the endogenous variables in detail.
Female Share of the Labour Force (FLF)
The female share of the labour force shows the extent to which women are active in the labour force. It is defined as the percentage of women in the total labour force, where the total labour force comprises men and women aged 15 and older who meet the ILO definition of the economically active population. It includes employed and self-employed workers as well as unemployed workers and first-time job-seekers. Furthermore, labour force measures generally include the paid workers in the agricultural sector, armed forces and seasonal or part-time workers. Yet it excludes homemakers, unpaid caregivers and workers in informal sectors (c.f. Morrisson and Jütting, 2005).
Data on the female share of the labour force are drawn from the World Bank’s World Development Index Data Base (2006) and cover the years 1980 to 2004 for 186 countries. The 4668 observations are distributed quite evenly across years and across countries. Early observations and observations of developing countries are adequately represented.
The female share of the labour force as percentage of the total labour force varies between 5,05% and 54,04% with a mean of 38, 70% over all countries and all years (see table 10).19 In 1980, the United Arab Emirates had the lowest female share of the labour force with 5,05%, and Cambodia had the highest with 53,23%. In 2004, West Bank and Gaza had the lowest female share of the labour force with 13,10% and Mozambique had the highest with 53,52%. The mean rose from 37,14% in 1980 to 40,24% in 2004, which shows that the range of the variable does not change very much over the 24 years.
Regarding only OECD countries, the values vary between 25,13% and 48,02%, with a mean of 41,29% over all years. In 1980, Ireland had the lowest female labour force with 27,96%, and the Czech Republic had the highest with 47,14%.20 In 2004, Turkey had the lowest female share of the labour force with 26,43% and Finland had the highest with 43,64%. The mean of the female share of the labour force for all OECD countries rose from 38,90% in 1980 to 43% in 2004. Hence, the range of the variable does not change very much over the 24 years in OECD countries, either.
19
Female Activity Rate (FAR)
The female activity rate is defined as the share of female formal sector employees in the female working age population aged 15 and over (in percentage points). Data on the female activity rate are drawn from the ILO Laboursta Data Base (2007) and cover the years 1960 to 2005 for 171 countries. In comparison with the female share of the labour force (FLF), the time horizon concerning FAR-observations is almost twice as long, but instead of 4668 observations for the FLF there are only 1453 observations for the FAR. The observations are irregularly distributed over time; there are much fewer observations for the 1960s and 1970s than from 1980 on. Moreover, Sub-Saharan Africa countries are strongly underrepresented.
The female activity rate as a percentage of the female working age population varies between 2,50% and 93,10% with a mean of 42,19% over all countries and all years (see table 10).21 In 1960, Egypt had the lowest female activity rate with 5,20%, and Thailand had the highest with 81,40%. In 2005, Tunisia had the lowest female activity rate with 24,40% and Ethiopia had the highest with 78,80%. The mean rose from 32,53% in 1960 to 49,00% in 2005. The range of the variable changes much more in comparison to the female share of the labour force because of the longer time period.
Regarding only OECD countries, the values vary between 17,00% and 74,20%, with a mean of 45,47% over all years. In 1960, Portugal had the lowest female activity rate (17,00%), and Turkey had the highest (65,40%). Turkey’s female activity rate had constantly fallen over the whole sample period and in 2005, Turkey denotes the lowest female activity rate in the OECD (24,8%). Denmark is on top of the list with 73,90%.22 The mean of the female activity rate for all OECD countries rose considerably from 35,59% in 1960 to 50,8% in 2005.
When interpreting data on female labour market participation, one has to distinguish between the female share of the labour force (FLF) and the female activity rate (FAR). These two indices of female labour market participation cannot be compared to each other. As the female share of the labour force (FLF) measures the share of women in the total labour force, its mean (38,65%) is naturally lower than that of the female activity rate (42,13%), which measures the share of working women in the female working age population.
21
The male activity rate a percentage of the male working age population varies between 51,20% and 97,00% with a mean of 77,00%.
22
Ratio female activity rate /male activity rate (RAR)
Data on the ratio of the female to the male activity rate are also drawn from the ILO Laboursta Data Base (2007) and include the years 1960 to 2005 for 171 countries. Like observations of the FAR, observations of the RAR cover a much longer time horizon than observations of the FLF, but there are also only 1453 observations. The observations are irregularly distributed over time and there are again much fewer observations in the 1960s and 1970s than from 1980 on and Sub-Saharan Africa countries are strongly underrepresented.
The ratio female/male activity rates varies between 0,30 and 1,08 with a mean of 0,56 over all countries and all years (see table 10). In 1960, Egypt had the lowest ratio (0,06) and Thailand had the highest with 0,91. In 2005, Turkey had the lowest ratio with 0,34 and Denmark had the highest with 0,68. The mean grew from 0,38 in 1960 to 0,68 in 2005.
Regarding only OECD countries, the observations vary between 0,19 and 0,90 with a mean of 0,63 over all years. In 1960, Portugal had the lowest ratio (0,19), and Turkey had the highest (0,70). In 2005 Turkey had the lowest ratio (0,34) and Denmark had the highest (0,90). The mean of the ratio for all OECD countries rose from 0,41 in 1960 to 0,73 in 2005.