Female Employment, Earnings Inequality and Household Well-being: The case of urban Turkey



Yüklə 446,5 Kb.
səhifə1/5
tarix12.10.2018
ölçüsü446,5 Kb.
#73218
  1   2   3   4   5


Female Employment, Earnings Inequality and Household Well-being: The Case of Urban Turkey
Meltem Dayıoğlu
(corresponding author)
Department of Economics

Middle East Technical University

06531 Ankara, TURKEY

dmeltem@metu.edu.tr

and
Cem Başlevent
Department of Economics

İstanbul Bilgi University

Kuştepe, Şişli

80310 İstanbul, TURKEY

cbaslevent@bilgi.edu.tr

Female Employment, Earnings Inequality and Household Well-being:

The Case of Urban Turkey
Meltem Dayıoğlu
Department of Economics

Middle East Technical University

06531 Ankara, TURKEY

dmeltem@metu.edu.tr

and
Cem Başlevent
Department of Economics

İstanbul Bilgi University

Kuştepe, Şişli

80310 İstanbul, TURKEY



cbaslevent@bilgi.edu.tr


Abstract
(102 words)
This paper deals with the distribution of female earnings in the urban areas of Turkey and the impact of these earnings on household income inequality and well-being. Using survey data from 2003, we demonstrate that the earnings inequality among women is higher than among men. We distinguish between the earnings of (casual and regular) wage and salary workers and the self-employed (including employers). In our household level analysis, we quantify the contribution of women’s earnings to household income and household income inequality. Our results reveal that the direction of the impact of female earnings depends on how one goes about measuring it.

1. Introduction
One of the salient features of the Turkish labor market is the considerably lower participation rates of women compared to their counterparts in the Western world. Recent estimates put women’s participation rate at 18.5% in urban and 39% in rural areas.1 While women’s participation in urban areas has remained roughly the same over the past two decades, there has been a continual decline in rural areas. The on-going structural change, which has resulted in the decline of the rural sector, is offered as an explanation for the declining participation rates of women. Traditionally, women have been employed in agriculture as unpaid family workers so that the move to urban areas has resulted in their withdrawal from the labor market. The relatively lower average education level of women vis-à-vis men is an important factor that hampers their participation in the urban labor market. However, the traditional division of labor where women’s proper role is seen as home-making is probably even more important in explaining the lower participation of women since, despite the recent improvements in demographics and educational outcomes, their labor market participation in urban areas has not recorded a significant increase. On the demand side, the relatively low employment generation vis-à-vis the growth in urban population, as well as the limited availability of part-time jobs and affordable child care facilities are also likely to have played a role in limiting women’s participation.
Female labor supply has been thoroughly studied in Turkey (see for instance Başlevent and Onaran, 2003, 2004; Çağatay and Berik, 1990; Çınar, 1994; Dayıoğlu, 2000; Özler, 2000; Tunalı, 1997, 2003; Tunalı and Ercan, 1998; Tunalı and Başlevent, 2005). There has also been work on the gender earnings gap (see for instance Dayıoğlu and Tunalı, 2003; Kasnakoğlu and Dayıoğlu, 1997; Tansel, 1994, 2005; Selim and İlkkaracan, 2002). However, the heterogeneous nature of the female workforce and its implications for individual and household welfare has not been the central issue in these studies. A preliminary analysis of the available data reveals that the earnings inequality is higher among working women than men, and the contribution of women to the household budget varies considerably depending on the nature of their participation. Our aim in this paper is to investigate this heterogeneity and how it relates to household income inequality and welfare. For this purpose, we make use of an official household budget survey (HBS) where both labor supply and income information are available. We have limited the study to the urban sector since this is where the diversity is really an issue. Another reason for this choice is that estimating the value of women’s work in the rural sector where it typically takes the form of unpaid family work is itself a major challenge.
The paper will contribute to the literature on the labor supply of Turkish women on two accounts: First, by drawing attention to the effect of women’s labor supply on household income inequality and welfare, which has not been done in Turkey. Secondly, by illustrating the inappropriateness of talking about a homogenous labor force especially in the case of women. By the end of this paper, we hope to have driven home the point that policies that aim to increase the participation of women need to take into account the diversity observed among both the potential labor market participants as well as the currently employed. We also hope that the findings of the study can help identify possible areas of intervention that can enhance the welfare of working women and their families and increase their labor market participation, which is recognized as one of priorities of Turkey in the process of accession to the EU.
The paper is organized as follows: The next section is devoted to a brief description of the labor market in the urban areas of Turkey. In Section 3, we present the data set and the methodology employed. In Section 4, we group women into regular wage earners, casual wage workers, self-employed, employers, and unpaid family workers and try to find a set of characteristics that distinguishes between these categories. In Section 5, we carry out a household level analysis where working women’s contribution to the household income is assessed. In Section 6, we discuss how women’s income alters household income inequality. Section 7 concludes the paper by offering a set of policy recommendations.
2. A Brief Account of the Labor Market in Urban Turkey
According to most recent (August 2006) estimates, Turkey’s population stands near 73 million with two thirds of the population residing in urban locations. Despite the decline in the population growth rate, the ongoing migration from rural to urban areas implies hundreds of thousands of newcomers to the urban labor market every year. The urban unemployment rate of 11.8% (vs. 5.3% in rural areas), despite a labor force participation rate of 49.5% which is well below the EU and the OECD averages (SIS, 2006). A closer look at the participation rates reveals that while men’s participation estimated at 72.9% (for the country at large) is close to the EU and the OECD average, women’s participation rate of 26.5% is less than half the EU and the OECD figures of 63.5% and 60.4%, respectively (OECD, 2006). While there is not much of a difference between urban and rural areas in regards to men’s participation, this is not the case for women. Furthermore, urban women suffer from higher rates of unemployment, which though not unique to Turkey, probably act as yet another factor discouraging women from joining the labor market.
Another prominent feature of the Turkish labor market is that many forms of employment exist, which in part serves the different needs of the market, but at the same time works to determine individual as well as household welfare. The distribution of employed men and women (ages 15-64) into various employment categories indicates rather different patterns. The main difference between men and women emerges in the frequencies of the ‘self-employment’ and ‘employer’ categories (see Table 1). While women constitute roughly 10% of the former, their representation in the latter is limited to less than 5%. In contrast, they are over-represented in the category of ‘unpaid family workers’, constituting 45% of that group.
--- Insert Table 1 about here. ---
There might be a whole host of factors that determine this distribution. Studies on the determinants of women’s participation point out that among other variables, education and marital status stand out as important determinants. Years of schooling can limit opportunities especially as regular wage earners since certain types of employment such as public sector jobs require at least a basic education diploma. Marital status, on the other hand, is indicative of the demand for women’s time at home. Married women’s participation is found to be the lowest among all demographic groups primarily because of existence of children and the traditional role attributed to women as home-makers. It is likely that the factors that determine women’s participation also determine their sector of choice. Although this study is not about the determinants of this distribution but rather its effects on individual and household well-being, it is still important that we try to understand the profile of women and that of their households by major employment categories. Table 2 provides such a description using the HBS data set (to be described in the next section) that we work with throughout the paper. The table also includes the corresponding figures for non-working women for comparison.
--- Insert Table 2 about here. ---
Among the distinct patterns that the subsample means reveal is that regular wage employment is typically held by younger and more educated single women. They come from smaller families with similar household incomes as that of non-working women. Employers, on the other hand, are the most educated group. They have smaller families, and substantially higher household income than other working and non-working women. What seems to be the main divide between the self-employed women and employers is their level of education. Self-employed women are older than employers but have substantially lower levels of education. They also come from larger households with more dependents and substantially lower household incomes. The lowest household income figures are observed for casual workers, who are the least educated group. The proportion of household income originating from non-labor income for this group is minimal. Their household sizes are not particularly higher but they have more children per woman. The unpaid family category also consists of relatively less educated women from larger families. The majority of adult males in such households are in the family business, which happens to generate household incomes substantially higher than most groups.
The data set also provides data on hours of work.2 In terms of the effort put into their jobs and remuneration obtained from them, casual workers fair the worst. They put in fewer hours over the year, and are therefore expected to have lower annual earnings. When their earnings are corrected by the number of hours, they still end up with substantially lower hourly earnings than any other group. Regular wage earners put in 46 hours per week on average which is what is expected of them given that the statutory work-week in Turkey is 45 hours. Employers work even longer hours than wage earners. The hourly earnings of the self-employed are substantially lower than employers due to the vast differences in human and possibly, physical capital they employ.
--- Insert Table 3 about here. ---
The distribution of working women into the various forms of employment also differs by sector of employment. In agriculture, we mostly find unpaid family workers and casual workers (See Table 3). Manufacturing employs a wide spectrum of women, though it serves as the main sector of employment for regular wage earners and the self-employed. Sales is another category that includes women of diverse backgrounds, though again it seems that employer women find this sector particularly open to them, followed by unpaid family workers, self-employed and finally, the regular wage earners. Other sectors where employer women are found include health services, other social and community services and to a lesser extend education services, manufacturing and real estate. The group that has a more balanced distribution across sectors is regular wage earners. Even among this group almost three quarters are found in the following four groups: manufacturing, sales, education and health services, which are the typical sectors where working women are found worldwide. As the figures show, the sector of employment pretty much goes hand-in-hand with the form of employment and within the employment boundary set by the society as appropriate for women. As noted above, we suspect that women’s educational attainment, demands on women’s time at home, the need for women’s monetary contribution, and the social values held by the community define the work opportunities of women. In our empirical work, we go beyond the sub-sample averages to illustrate this divide among working women, and its impact on household well-being and income distribution.
3. Data and Methodology
In the empirical work, we use data drawn from the 2003 Household Budget Survey (HBS) conducted by the State Institute of Statistics (SIS) of Turkey. The main reason for the use of this survey rather than the Household Labor Force Survey is that the latter lacks information on individual and household income. The 2003 HBS covers 73,032 individuals from 18,278 urban households. From this, 16,742 households where there is at least one woman between the ages of 15-64 are drawn. Since one of our interests in the paper lies in determining the contribution of women to household income, we have further excluded 1,152 one-person female-headed households.
Not being the product of a labor force survey, the data we use in this study define employment over a longer reference period than one week which is the standard for the former. The 2003 survey inquires whether the individual is employed in the month that the survey was carried out and at any time in the 12 months preceding the survey. The information relating to the form of employment was collected in relation to the current job held. For individuals who were not employed in the survey month but who worked for some time during the past year, information on employment status is available through the last job held. The ‘income’ variable includes after tax, in-cash and in-kind payments resulting from primary and secondary jobs as well as from non-labor sources, such as interest and rent incomes, dividends, transfers and the like. For home-owners, this source also includes self-reported imputed rents. The survey reports both the monthly and annual incomes of individuals. However, the former might not reflect the true standing of the household in the income distribution if the flow of income is irregular. For this reason, we have opted to rely on annual incomes. Given our emphasis on the various types of employment, we compare the earnings of individuals from wage-and-salary and own-account work. Where necessary, we impute wages for unpaid family workers. We do this by estimating what their hourly earnings would have been had they chosen wage employment. In doing so, we rely on estimated parameters from an the Mincerian earnings function that involves correction for selectivity using the Heckman (full-information maximum-likelihood) procedure. The predicted hourly wages for these individuals are multiplied by their annual hours of work to obtain annual earnings.
One of our main goals in this paper is to illustrate the diversity among employed women. In the preceding section, we have done this partly on the basis of average individual and household socio-economic characteristics. We have also seen that the end result of the chosen work status is to produce substantial earnings differentials among women. In the following sections, we investigate the earnings differentials in more detail by looking at the overall distribution in each category and by using the various measures that summarize the distribution that are more informative than simple averages. Specifically, we measure earnings inequality using the Gini coefficient and the squared coefficient of variation (SCV), which are among the most commonly used measures of income inequality. As the basis of our comparisons, we use the annual earnings of individuals. As illustrated in Table 1, annual hours of work varies with the form of employment, so that part of the variation in annual earnings must stem from hours of work. To be in line with the household level analysis and assuming that the chosen form of employment implies certain work hours and is remunerated accordingly, we carry out the individual level analysis on annual rather than monthly or hourly earnings.
The second major goal of this study is to evaluate the contribution of women to household welfare and the impact of their earnings on overall household income inequality. The former is a complex issue requiring information on the relative power sharing within the household and the resulting intra-household distribution of resources. Although we do recognize that income is not an end in itself, especially in a social context where traditionally labor market is regarded as off-limits for women, we nevertheless measure women’s contribution simply via her earnings and consider it as her potential contribution to household well-being. In this exercise, the second assumption we make is that regardless of whether women join the labor market or not, their contribution to household well-being through re-productive work is the same. This assumption is based on a study by Kasnakoğlu and Dayıoğlu (2002) who find that the hours of work put into household chores are marginally greater for women who work in the labor market.3 This is not surprising given the fact that men contribute very little to household chores and goes to indicate that women who join the labor market do double shifts.
In assessing the relative contribution of women to household income, we observe the relevant figures by income quintiles constructed on the basis of household incomes corrected by the Eurostat adult equivalence scale as well as the consumer price index figures available at the province level. The former correction is done mainly to account for economies of scale within the household. Instead of merely counting the number of household members the Eurostat scale distinguishes between adults and children, where the first adult is counted as 1, each additional adult as 0.5, and children (i.e. ages less than 14 years) as 0.3 adults. The latter correction is necessary to account for across region variation in cost-of-living.
In determining whether women’s earnings are ‘equalizing’ or ‘disequalizing’ with regard to the overall income inequality, a number of strategies can be adopted. In fact, during the past four decades, an extensive literature on the measurement of the contribution of income sources to inequality in total income has developed in an effort to properly to account for the impacts of the ‘factor components’ under examination.4 The need for such decomposition methods arose mainly as a result of the recognition that the more simplistic ‘before-after’ analyses – which look at the change in inequality when a component is added in to income figures – were likely to produce misleading conclusions. It is now well-known that which methodology is appropriate, and thus whether a source is ‘equalizing’ or ‘disequalizing’, may depend on which question one is asking. As summarized in Lerman (1999), “an investigator may have an interest in the change in inequality resulting from: (1) a marginal change in an income source, (2) an income source becoming equally distributed, (3) the elimination of an income source, (4) a distribution in which the income source becomes the only source of inequality, and (5) changes in rankings due to changing income sources.”
In our computations, we primarily rely on Jenkins’s application of the Shorrocks (1982) formulation which computes the proportionate contribution of each source. In terms of Lerman’s categorization, this method corresponds to the average of items (2) and (4). The Shorrocks formulation is based on the covariances between the values of the factors and total income, and it is independent of the choice of the measure of inequality. Jenkins (1995) as shown that if the coefficient of variation (CV)5 is used as the inequality measure, proportionate contributions can be written in terms of (i) factors’ correlations with total income, (ii) factors’ shares in total income, and (iii) factor inequalities measured by the CV. The proportionate contribution of factor Fk is given by
ρk × [mean(Fk) / mean(total)] × [CV(Fk) / CV(total)],
where ρk is the correlation coefficient between Fk and total income.6 We also carry out a ‘before-after’ type analysis to measure the sensitivity of our findings to a change in methodology. This involves measuring the inequality in household incomes by first excluding and then, including female earnings. This procedure corresponds to Lerman’s third category.

4. Diversity and Inequality
The descriptive statistics presented earlier have shown that there is appreciable variation in the average individual and household characteristics of working women which suggests that women with quite diverse backgrounds enter the labor market. We have also shown that the employment outcomes of women in terms of earnings differ depending on the status chosen. In this section, we further illustrate this diversity among working women based on their earnings by employment status and by comparing these figures with the earnings inequality observed among men.
--- Insert Figure 1 about here. ---
In line with the literature on gender inequality in earnings, we find that women’s annual earnings are considerably lower than men’s. The average female wage is 65.1% of the average male wage. The empirical cumulative distribution functions shown in Figure 1 illustrate that women earn less than men in all forms of employment and the earnings gap is greatest at the bottom of the respective distributions. Looking at the issue from a different perspective, though women workers (excluding unpaid family workers) constitute 19.6% of the workforce, they make up 36.7% of the bottom 20% when grouped in terms of annual earnings. In contrast, they constitute 10.5% of the top 20%.7 Examining the two extreme groups in more detail from the perspective of women, we observe that the top group is mainly composed of regular wage earners and employers. Although the overall share of the wage earners is 73.1% among female workers who are gainfully employed, their representation in the top quintile reaches 87.3% (see Table 4). Employers constitute only 1.3% of female workers, but make up 5.6% of those in the top 20%. In contrast, in the bottom 20% of the distribution we see an over-representation of causal workers and the self-employed. While the former makes up 18.2% of the working female population, they constitute 37.7% of the women who are at the bottom 20% of the earnings distribution.
--- Insert Table 4 about here. ---
The distributional statistics presented in Table 5 indicate that the earnings of women are considerably more unequally distributed than those of men. While the Gini coefficient is 0.45 for men’s earnings, it is on the order of 0.52 for women (excluding unpaid family workers). When the distributional statistics are examined by employment status, higher inequality figures are obtained for women in all categories. Among women, the group that displays the highest inequality is the self-employed, followed closely by employers. In the case of men, this order is reversed; the most heterogeneous group being employers, followed by the self-employed.
--- Insert Table 5 about here. ---
Another interesting exercise is to compare the incomes of women belonging to the same distributional rank across the employment categories. For instance, the median earner working as self-employed receives 34% of what a regular wage worker in a similar position earns. The gap opens drastically when the earnings of two women at the bottom 10% of their respective distributions are compared; the self-employed women receive, on average, only 19% of the earnings of regular wage earners. The figures provided in italics in Table 5 compare the earnings of different groups of men and women at various positions in their respective distributions with the earnings obtained for that decile disaggregated by gender. The results show that the earnings inequality at the bottom of the distribution is higher than what is observed at the top. While the female regular wage earner at the bottom decile earns 1.6 times the (weighted) group average, the corresponding employer earns 6.6 times this amount, whereas the causal worker earns less than a half and the self-employed about a third. At the top decile, the distribution improves somewhat, with the earnings of regular wage earners steadily approaching the mean from the top and the self-employed from the bottom. The earnings gap between the employers and the other groups somewhat closes, and becomes 3 to 4 times the group average. Causal workers, on the other hand, fair worse in upper quintiles. It is interesting to note that the self-employed women at the top 10% receive over three quarters of the earnings of top earning regular wage earners. These findings are indicative of the particularly sharp diversity observed among the self-employed. A quick look at the pattern observed among men shows that the within group diversity is not as large as what is observed among women. Broadly speaking, regular wage earners and the self-employed earn amounts close to the mean for their group, causal workers half, and the employers 2 to 2.5 times this figure.
The distributional results indicate that working women not only differ among employment categories but even sharper differences exist among women categorized under the same employment category. As discussed earlier, a whole host of factors are possibly at play here, among which schooling is a prime cause. Looking at the employment categories by education does provide hints as to why certain groups of women display larger variation in earnings (see Table 6). Practically speaking, regular wage employment is closed for women with little or no education. The majority of these women end up working as casual workers. Similar observations are made for men as well. The employer category, on the other hand, includes highly educated women. In fact, over half of this group is made up of women with higher education. Another 23% have a high school diploma. Amidst this highly educated group (characterized by doctors, lawyers, professors and the like) there are also those with no or very little education, primarily engaged in manufacturing and agriculture. The self-employed consist of an even more diverse group though the diversity is brought about by a small number of highly educated women.
--- Insert Table 6 about here. ---
Interesting results also emerge when the educational structure of these groups are contrasted with the structure observed for men. As is the case for women, the proportion of highly educated men joining the labor market as self-employed is rather small. However, while about 20% of the self-employed women have no or minimal education, this figure is less than 6% for men. In fact, in all categories under examination, relatively fewer uneducated men are found, despite the fact that the average education level of working women (8.7 years) surpasses that of men (8.2 years). This finding stems from the fact that the female population is relatively less educated; while 10.6% of the female population is illiterate, the corresponding figure among men is 1.6%. The fact that employer women are considerably more educated shows that this type of employment is open to a certain group of women in sectors that require high levels of schooling such as medicine, law, teaching and the like. The proportion of highly educated female employers in sectors such as manufacturing (where 21% of highly educated men are found) is nil. The data show that an average business run by such women employs 3.6 persons.
The diversity observed among working women is expected to affect the contribution they make to household income, and therefore, play a role in determining the position of their household in the overall income distribution. This is what we investigate next.

Yüklə 446,5 Kb.

Dostları ilə paylaş:
  1   2   3   4   5




Verilənlər bazası müəlliflik hüququ ilə müdafiə olunur ©genderi.org 2024
rəhbərliyinə müraciət

    Ana səhifə