A Comparative Study on Gender Discrimination and Its Indications: Principle and Facts from India

The beginning of any systematic women’s studies in India is usually dated to the publication of the report ‘Towards Equality’ in 1974 (Government of India 1974). This document, by scanning available data, was able to establish the size and the significance of gender gaps in post-independence India, and stimulated a wide spectrum of both analytical and empirical work. It is customary to start any discussion on gender by distinguishing between the terms ‘sex’ and ‘gender’. ‘Sex’ being a biologically determined attribute is not easy to change while ‘gender’ refers to socially constructed roles, which are likely to vary from one society to another, and which change significantly as societies develop and evolve over time. The construction of gender roles in any society is influenced by several factors, most of which are largely independent of the sources of poverty or growth. An important contribution of the report ‘Towards Equality’ was its clear articulation of the two sets of factors that play a role in the construction of gender in a society. These two sets of factors are those that determine ‘achieved status’ and those that determine ‘ascriptive status. Achieved status is ‘a status acquired by an individual through her efforts, often through competition and the use of special abilities, knowledge and skill.’ Ascriptive status refers to ‘any status not based on individual ability, skill, efforts or accomplishment but on inherited positions in society’ (Govt. of India 1974 : 7)2. “Achieved status’ is determined by such attributes as education, health, work participation and so on, essentially characterized by the fact that these are individual attributes, and are amenable to change through individual decision. In contrast, ‘ascriptive’ status include factors that are not in the control of the religion etc. The basic motivation behind research in this area is to identify the best points of intervention that would help to bring about change in gender relations, in the direction of greater equality, in a context of economic growth and human development. Development, in the simple sense of increasing material welfare, has dominated discussion and planning for growth since the 50’s, although the strategies recommended have changed considerably over the years. But it is only since the 70’s that gender equality has been an objective in the formulation of national macro policy, and this owes much to the advocacy and intervention of international bodies, specially the UN system. Clearly, one can think of gender equality (‘equal access and opportunities for men and women’) as being realized only within an enabling framework of equal rights and equal access. The legal and constitutional framework of rights therefore forms an essential component in any discussion. It can be argued that economic equality is needed to make other equalities real. Indicators frequently used to measure gender equality include life expectancy, education, and earnings. These give an indication of the degree of access to food, health, education, and the means of livelihood. Other indicators are sometimes suggested to give a more direct measure of the extent to which women have control over their lives: one is the degree of seclusion expected from women. Interpretation of change is equally complex (see Saith and Harris-White 1998). Gender discrimination against women in the market place reduces the available talent in an economy, which has negative economic consequences. Gender discrimination takes many forms. Many social practices seen as normal from a religious or cultural point of view (which may have deep historical roots) leave women out of the economic mainstream. These social practices may have profound economic consequences because they do not allow society to take advantage of the talent inherent in women. This paper investigates these economic consequences. Although gender discrimination may have a myriad of other important consequences, including psychological, sociological, and religious, these are not discussed in this paper. We develop a theoretical model that allows us to explore the economic implications of gender discrimination in the labor market. In the model, individuals are born with a given endowment of entrepreneurial talent and decide how much human capital to acquire, and whether to become managers or workers. Individuals can also engage in home production. Their choices depend on what everyone We study three possible scenarios. First, we analyze the labor market equilibrium without discrimination. Second, we model gender discrimination as an exogenous exclusion of females from managerial positions. Our model shows how this discriminatory practice affects the labor market, the equilibrium wage rate, the allocation of talent across working and managerial positions, the investment in education by individuals (males and females), and economic growth. We show that discrimination tends to lower equilibrium wages for female and male workers, and to reduce investment in human capital by all females and by male workers. We also show that the average talent of managers is smaller in case of discrimination, which accounts for reduced innovation in the economy, and that the average productivity of workers, which accounts for technology adoption in the economy, is reduced too. Both factors lower economic growth. The fact that the relative average earnings of females relative to males are lower due to occupational segregation matches empirical evidence about the gender wage gap. While gender discrimination against women in the labor market in developed countries is usually identified with differential wage rates, it is in developing countries that this discrimination appears to take the form of differential access to wage employment (Collier (1994)). To test the implications of the model we take the particular case of India. This is a relevant case for the current study since cultural reasons are there (as in many other countries) known to restrict women’s access to work (Kumar et al. (1999)).

Dreze and Sen (1995) argued that “the persistence of extraordinarily high levels of gender inequality and female deprivation are among India’s most serious social failures. Few other regions in the world have achieved so little in promoting gender justice” (p. 177-178). Based on a nationally representative sample survey of about 90,000 women (IIPS and ORC Macro 2000), it was found that a majority of women in India married before they reached the legally prescribed minimum age of 18 years. This included 41 per cent of urban women, much against the spirit of the Child Marriage Restraint Act of 1978. The survey also indicated that two thirds of women were not regularly exposed to any mass media, including 13 percent of the women living in urban areas. Labour force participation can be a preliminary step towards women’s empowerment, although this effect depends upon the type of occupation, amount of earning, and the women’s actual control over

women, according to Sen (1990). However, it was found in the above survey that only 24 per cent of urban women were gainfully employed. One-fifth of that group reported that their family was entirely dependent on their earnings. Another 30 per cent stated that they contributed about half of the total family earnings. The survey made some notable revelations about women’s control over their finances. Thirty per cent of the women who earned money said that their husbands and others in the household made all the decisions on how the money they earned would be used, while 57 per cent of the urban women who were earners said that they made some decisions on their own regarding how to spend the money they earned. Just working at a paid job did not actually result in greater autonomy and freedom within the household for most women. One factor leading to strong disincentives and discrimination against daughters is the practice of dowry, which together with marriage costs is a major drain on household resources (Menski, 1998). Dowry demand is responsible for many ills perpetrated against women in most parts of India. Both the emergence of dowry in many communities that did not traditionally practice it, and a phenomenal increase in marriage expenses, influence parental decisions to get rid of daughters. That is why in many Indian families the birth of a son is an occasion for celebration and the birth of a daughter is often considered as a time of crisis. Though many argue that with the increase in education and economic opportunities son preference will eventually decline, evidence from recent studies indicates, to the contrary, a strengthening of gender bias in low fertility areas, even when education and income are improving (Das Gupta and Bhat, 1997).

Indicators of gender discrimination seek to go beyond description, and to identify policy measures for improving women's status or autonomy. These terms are not synonymous. Status has the connotation of relative social standing, and improving status may not increase autonomy, a term that suggests the ability of self-determination, independence and control over one's life. However, knowledge of status is important in defining norms of behavior, and permissible deviations from such norms. Demographic studies have tried to explore the extent to which women have control over their fertility behavior by using various measures of ‘autonomy’. interesting dimension of autonomy, draws attention to the difference between autonomy that is culturally `granted' and that which is sought. Most of the measures above reflect a social consciousness rather than individuality. Women do not `choose' where to have their children, how long to stay in school, or when and whom to marry. In fact, real autonomy in a majority of Indian households may be closely related to age, so that in a very real sense older women have the power to `give' greater autonomy to young women. As Basu puts it, `The real pity often is not that men wield so much domestic power, it is that during the prime reproductive years female power is at its lowest' (Basu 1996 : 54). `The `socialisees' rather than the already `socialized' represent the potential agents of change' (Epstein 1982) Education and Employment: Education and employment are undoubtedly the most popular choices of ways to improve women's well being. The cause of women's education, in particular, has received much support from the findings of demographers. `A number of empirical studies indicate that the extent of anti-female bias in survival is substantially reduced by various influences that give women more voice and agency within the family. One of these influences is female education. Another is the ability to earn an independent income through paid employment.' Employment as a route to empowerment is equally complex. Work participation levels of women are high, if an extended labour force definition is used, although the majority are in informal sector jobs, crowded into the low skill end of the spectrum, and usually in part time work. The uncertain impact of paid work on women's welfare is closely related to their continued home responsibilities. Does earning an independent income increase a woman's bargaining power? The answer is yes if she has real control over it. In many situations however women work in response to household needs, and have been variously described as `target earners' (Mukhopadhyaya 1995) or as a `flexible resource of the household' (Banerjee 1998`). It is unlikely that earning an income will alter the balance of power substantially; more likely, it will reflect the pre -existing balance of power in a household. If this is reasonably good (a complex of individual and cultural attributes) work may be empowering; if bad, it can lead to higher levels of stress and continued exploitation. Demographic Transition and Access to Resources : While demographic transition has accompanied, or Son preference that characterizes India, and Asia in general, does not disappear with rising incomes. The discrimination evident in a declining sex ratio and in the large numbers of `missing women' suggests that there is no reduction of gender bias in a context of economic growth (Dasgupta and Bhat 1998). Indicators reflecting the health status of women show the same patterns. First, the perception of poor health itself varies across gender. Second, there is some evidence that medical intervention is sought more often for boys than girls. Third, there is very inadequate allocation for health needs special to women. The Cairo Conference on Population and Development (1994) recognized the need to address health and demographic concerns in a broader framework. The programme of action recognizes the need for sustained economic growth in the context of sustainable development; education, especially for girls; gender equity and equality; infant, child and maternal mortality reduction; and the provision of universal access to reproductive health services, including family planning and sexual health.

The type of discrimination used in this paper is related to the concept of discrimination on grounds of employers’ tastes, which was first used by Becker (1971), and may be rational in the context of religious or traditional beliefs that may operate as social norms in many countries.8 The concept of social norm that we use as to explain what we call total discrimination is related to the concept of social stigma in Goldin (1994). We consider discrimination as exogenous while there may certainly be a bunch of factors that can account for discrimination up to some degree. However, we do explore the sustainability of the operating social norm. We argue that inefficiencies arise due to distortions in the allocation of talent. The idea that distortions in the allocation of talent across occupations or sectors have negative growth implications is not new (Murphy et al. (1991), Fershtman et al. (1996)), but to our knowledge this is the first study to use it in order to analyze the consequences of discrimination. Since the mid-1990s, newer decomposition methods have gone beyond the focus on single summary measures. Studies now seek to go beyond the mean and answer questions like what happened where in the distribution, rather than decomposing only differences at the means. Additionally, since policies could have different effects at different parts of the wage distribution, this nuanced understanding is needed to assess their impact on wage inequalities. For example, a change in minimum wage regulations is more likely to influence the lower part of the wage distribution. quantiles of the distribution. Thus, Jenkins (1994) criticizes the “almost universal practice to analyze discrimination at the average proportionate wage differential” (p. 82). While the summary scalar measure of discrimination from the B-O decomposition of wage inequality may be the same for two populations, it is possible that the distribution of discrimination across different parts of the wage distribution may be very different for the two. As with the B-O decomposition, the starting point of these decompositions is the creation of a counterfactual. The methods which study wage gaps throughout the wage distribution rely on estimation of counterfactual wage distributions using various methods. Albrecht et al. (2003) was the first paper to introduce a concrete way of defining the glass ceiling phenomenon in wages. They interpret a glass ceiling as an increasing gap in log wages of two groups as one moves from lower to higher quantiles, with a sharp acceleration at the upper tail of the distribution. They found strong evidence of the emergence of a glass ceiling in Sweden, especially in the 1990s. They also found that the glass ceiling effect persists even when controlling for individual differences in characteristics such as age, education, industry/sector of employment and immigrant status. As expected, including detailed occupational controls did lead to decline of the wage gap at the top of the distribution indicating a dampening of the glass ceiling effect. Machado and Mata (2005) put the quantile regression (QR) technique to great use in assessing how different variables affect inequality. For instance, how higher education affects the top of the distribution may be very different from how it affects the bottom and this has implications for influencing overall wage inequality. This is where the QR technique can be far more informative than the Ordinary Least Squares (OLS) regression which only looks at mean effects. The Machado Mata (MM) decomposition splits the changing wage inequalities in Portugal over the period 1986-1995 into two components, one representing the changing characteristics and the other the changing returns to them. Machado and Mata (2005) put the quantile regression (QR) technique to great use in assessing how different variables affect inequality. For instance, how higher education affects the top of the distribution may be very different from how it affects the bottom and this has implications for influencing overall wage inequality. This is where the QR technique can be far more informative than the Ordinary Least Squares (OLS) regression which only looks at mean effects. The Machado Mata (MM) decomposition splits the changing wage inequalities in Portugal over the period 1986-1995 into two components, one representing the changing characteristics and the other the changing returns to them.

1995-2001. They found considerable variations in the raw gender wage gap between European nations. They estimated the quantile regressions separately for three sub-samples (public sector, private sector and both combined) for all countries. Like most studies estimating wage equations, the analysis is carried out including industry and occupation in one model and excluding them in the other specification. Using MM decompositions, they conclude that for all countries and across all quantiles, gender wage gaps exist even when men and women are assumed to have the same characteristics. Chi and Li (2008) use the RIF decomposition technique to study gender wage differences in urban China from 1987-2004 and find that the sticky floor effect is present in China as opposed to the glass ceiling and that this effect has become stronger over the time period under the study. This effect is mainly responsible for the increasing mean gender wage gaps in China. Both the composition effects as well as the structure effect (unexplained/discrimination component) have increased over time.

This paper provides theoretical and empirical support for the view that gender discrimination acts as a brake on economic development. We find that a 10 percent increase in the female-to-male managers ratio in India would increase total output per capita by 2 percent, while a 10 percent increase in the female-to-male workers ratio would increase total output per capita by 8 percent. This is in accord with the theoretical predictions of our model, according to which gender discrimination in either managerial positions or the overall labor market has negative economic consequences. The intuition is as follows. If women cannot gain access to managerial positions, the equilibrium wage rate declines, and the cutoff level of talent of managers declines as well, so that the average talent of entrepreneurs and economic growth both decline. If females cannot participate in the labor market, but have to engage in home production, the wage rate is the same as without discrimination, so that the cutoff level of talent is the same and, therefore, there are no innovation or adoption implications. Nevertheless, growth decreases due to the fact that half the population acquires zero education. Moreover, the theory predicts that per capita output is lower than it is without discrimination. What transpires from the analysis above is that measuring gender discrimination is no easy task. The problem arises not simply because the phenomenon roughly the same direction under all conditions. Thus determining the context – specificity of different indicators becomes important, as is the need for establishing systematic patterns of linkages between them for informed analysis. Mapping these complex sets of issues against the background of changing economic policy regims poses another set of challenges. Changes in macro policy would generally get manifested in changes in external economic opportunities and constraint for households at various levels. These may trigger gender-differentiated changes in labour use patterns, both within and outside the households, generate some new pressure points and relieve some other. Tracing them systematically to track the changing nature of gender relations in a particular context is a challenge in itself.

• Becker, Gary S., 1971, The Economics of Discrimination (Chicago: The University of Chicago Press)

• Blinder, Alan (1973), “Wage Discrimination: Reduced Form and Structural Estimates, Journal of Human Resources 8:436-455

• Coate, Stephen, and Sharon Tennyson, 1992, “Labor Market Discrimination, Imperfect Information and Self Employment”, Oxford Economic Papers, 44, pp. 272-88.

• Croll, Elizabeth (2002). “Fertility Decline, Family Size and Female Discrimination: A Study of Reproductive Management in East and South Asia,” Asia Pacific Population Journal, 17(2), pp. 11-38.

• Das Gupta, Monica (1987). “Selective Discrimination against Female Children in Rural Punjab, India,” Population and Development Review, 13(1), pp. 77-100.

• Deaton, Angus, 1989, “Looking for Boy-Girl Discrimination in Household Expenditure Data”, World Bank Economic Review, 3 (1), pp. 1-15.

• Esteve-Volart, Berta, 2000, “Sex Discrimination and Growth”, Working Paper No. 84, International Monetary Fund

 Francois, Patrick, 1998, “Gender Discrimination without Gender Difference: Theory and Policy Responses”, Journal of Public Economics, 68, pp. 1-32.

• Madheswaran, S. & Khasnobis, B.G (2007) “ Gender Discrimination in the Labour Market: Evidence from the NSS”, WIDER research project on “Gender wage Gap and its Impact on poverty: Evidence from India”

 Oaxaca, Ronald L. & Ransom, Michael R. (1994) “On discrimination and the decomposition of wage differentials,” Journal of Econometrics, Elsevier, vol. 61(1), pages 5-21, March.