Gender Inequalities in Tasks and Instruction Opportunities within Indian Families

Abstract

This contribution uses the Indian Time Use Survey (ITUS 1999) to document gender inequalities in tasks in India and their impact on an important aspect of inequality of opportunity – the resources invested in the education of children. It examines the school attendance of Indian children and the probability that they receive informal instruction or assistance with learning at home. The analysis documents clear gender inequalities in the allocation of household tasks among girls and boys and their parents, but finds more mixed evidence regarding gender favoritism in human capital investment. As children living in rural areas grow older, school attendance falls off much more rapidly for girls than for boys; but in urban areas, attendance of boys and girls remains essentially similar. The paper estimates a household fixed-effects model of the probability that a child receives informal instruction at home, and finds no evidence of gender-based discrimination.

Keywords:

• Economic development
• education
• household behavior
• time use

ACKNOWLEDGMENTS

This paper would not have been possible without the very generous help of Professor Indira Hirway, Director and Professor of Economics, Centre for Development Alternatives, Ahmedabad, India. Professor Hirway was instrumental in the design of the ITUS, and her generosity in obtaining and sending us the micro data from this survey is deeply appreciated. We would also like to thank the guest editors of the volume, three anonymous referees, Nancy Folbre, Sue Himmelweit, Thomas Masterson, Killian Mullan, and other participants at the Canadian Economics Association meetings in Vancouver on June 6, 2008, the ASGE/IAFFE session on “Time Use, Unpaid Work, and Public Policy” at the ASSA Conference, San Francisco, January 5, 2009, and the March 9–10, 2009, Workshop at American University for their helpful comments. Indrani Namilikonda provided feedback both as an intelligent lay person and as a married woman residing in urban India. Remaining errors are our responsibility.

Notes

¹ F. Thomas Juster and Frank P. Stafford argue that “the fundamental scarce resource in the economy is the availability of human time” (1991: 471).

² The personal interview methodology was very labor-intensive, but was considered necessary to collect reliable diary data from respondents who are, in some cases, illiterate. Jonathan Gershuny (1999) discusses the advantages of the diary methodology, which walks the respondent sequentially through the previous day's activities, in improving recall and imposing aggregate consistency of responses. An “abnormal” day is defined in the Instruction Manual for the Field Staff (Government of India 1999: 36) as “that day of the week when guest arrives, any member of the household suddenly falls sick, any festival occurs, etc.” The “weekly variant” is “determined according to the pattern of the major earners' holiday. If the major earner does not holiday, then school children's holiday will be taken. If even this is not applicable, then day of weekly hat (bazaar) may be taken” (36).

³ Although one might wonder whether six states' data could fully capture the diversity of India, Hirway has argued that “cross-checking of the results has confirmed that the sample is fairly representative of the country” (2000: 11). In any event, this data would be interesting even if it were only seen as a sample of the approximately 233 million people inhabiting these states (Census of India 2001). The figure would be 254 million if we included the state of Chattisgarh, which was carved out of Madhya Pradesh in the year 2000 (after the ITUS fieldwork). These states are different in many ways. For example, the primary language in Orissa is Oriya, whereas in Tamil Nadu, it is Tamil. The socioeconomic composition (in terms of caste, religion, etc.) of these states is also different. For example, the percentage of Scheduled Tribes (described in detail below) in Meghalaya is higher than that in other states. The geographical terrain and ecological conditions that characterize these states are also different. For more details on these states, see Government of India 2010.

⁴ See also R. Baskaran (1999).

⁵ As one would expect, if we look at instances of paid activities, most of them fall under primary, secondary, or trade: 66.14, 13.94, and 19.81 percent, respectively, in rural areas; and 8.20, 23.40, and 68.37 percent, respectively, in urban areas. The rural-urban differences are also unsurprising.

⁶ Activity Codes: 5 – Care for Children, the Sick, Elderly and Disabled for Own Household; 6 – Community Services and Help to Other Households; 7 – Learning; 8 – Social and Cultural Activities, Mass Media, etc.; 9 – Personal Care and Self-Maintenance.

⁷ Specifically, Primary Production activities includes the major categories: 11. Crop farming, kitchen gardening, etc.; 12. Animal husbandry; 13. Fishing, forestry, horticulture, gardening; 14. Collection of fruit, water, plants, etc., storing and hunting; 15. Processing and storage; 16. Mining, quarrying, digging, cutting, etc. Under Secondary Activities, the major headings are: 21. Construction activities, and 22. Manufacturing activities – each with numerous subheadings. The Trade, Business and Services activities include: 31. Trade and Business, and 32. Services – which (to give readers an idea of the detail of classification) includes: 321. Service in government and semi-government organizations (salaried); 322. Service in private organizations (salaried); 323. Petty service: domestic servants, sweepers, washers, pujari, barber, cobbler, mali massaging, prostitution, watching and guarding; 324. Professional services: medical and educational services (private tuition, nonformal teaching, etc.), financial services and management and technical consultancy services; 325. Professional services: computer services, Xerox/photocopying services, beauty parlors, hair cutting saloons, etc.; 326. Technical services: plumbing, electrical, and electronic repair and maintenance, and other related services; 327. Others; and 329. Travel to work.

⁸ The primary activities, especially agriculture, display some seasonality, with the average times varying across months.

⁹ In one line of argument, “work” is an instrumental activity – one “works” in order to produce the goods and services that are inputs into utility, while care for others is meaningful in itself. Others see this distinction as negating the meaning and other “process” benefits that most people find in paid employment. On examining time-use surveys across the world (Indira Hirway 1999), we discovered that sometimes care activities are grouped with “voluntary” work. In the context of ITUS, care activities have been classified under “extended work” (Hirway 1999) or “non-market household production” (A. C. Kulshreshta and Gulab Singh 1999: 1). On the issues involved in classifying various kinds of work in the ITUS, see Hirway (1999, 2000).

¹⁰ Note that these averages are for all the boys and girls (both attending and not attending) and are therefore not reported in Tables 1b and 1a; they are computed separately.

¹¹ “Leisure” is an ambiguous category (Lars Osberg 2008), but the broad activity groups 8 (Social and cultural activities, mass media, etc.) and 9 (Personal care and self-maintenance) come closest – see, for example, activities 863. Reading newspaper and magazines; 852. Watching television and video; 951. Talking, gossiping, and quarreling, 961. Doing nothing, rest, and relaxation; not to mention 911. Sleep and (euphemistically described) related activities. Urban women over 45 years report substantially more such time than younger women, and somewhat more, on average, than men.

¹² There is a large literature on education of children in India. In the interest of space, we do not survey this literature here, but see PROBE Team (1999), Dreze and Sen (2002), and Motiram and Osberg (2007).

¹³ Although P. Duraisamy (2002) provides estimates of the rate of return to education in India between 1983 and 1994, and argues that the returns to girls' schooling in India typically exceed the rate of return for educating boys, James Heckman, Lance Lochner, and Petra Todd (2006) emphasize the complexities involved in providing an unambiguous estimate of “the” rate of return to years of education. Furthermore, the work in Dreze and Sen (2002), particularly chapter 5, is representative of a large literature that emphasizes the huge variance in quality of schooling in India and the low quality of much of the public school system.

¹⁴ As clarified later, we use the term “parental instruction” somewhat broadly to refer to instruction by all adults in the household.

¹⁵ In Scandinavia in the seventeenth century nearly universal literacy was achieved, as Egil Johannson notes, “almost completely without the aid of a proper school system in the countryside. The responsibility for teaching children to read was ultimately placed on parents and godfathers” (1988: 137). (Swedish parents and godparents took this responsibility seriously, because Lutherans then believed in the possibility – even the certainty – of eternal damnation of the souls of the children who did not learn their catechism before confirmation, typically at age 13 or 14). The religious roots of this early literacy imply that it was not caused by economic development – but it did facilitate/enable Scandinavia's later development.

¹⁶ “Household” is defined in the ITUS Instruction Manual (1999: 8) as: “A group of persons normally living together and taking food from a common kitchen… the members of a household may or may not be related by blood to one another.” The two common structures are a joint family and a nuclear family. Informal instruction can (and is) done by parents, but also by other members of the family (for example, a grandfather or elder brother).

¹⁷ One would expect that informal instruction affects and is affected by the school success of children. Unfortunately, the data do not allow us to investigate this phenomenon.

¹⁸ We cannot, for example, distinguish between the hypotheses that (a) 42 percent of rural households help with homework, but only for one day each week or that (b) 6 percent of rural households help with homework every day of the week.

¹⁹ The Indian caste system is quite complex and has been subject to much analysis by social scientists (see Dipankar Gupta 1993 and the references therein). The SCs are the formerly untouchable castes, and people belonging to them are sometimes referred to as Dalits. Although there is no single hierarchy that prevails throughout India, broadly speaking, the SCs are at the bottom of the caste hierarchy. The STs can be loosely thought of as indigenous groups. Both the SCs and STs have been historically subject to considerable discrimination, and the Indian constitution has therefore provided several measures to protect them. Despite this, they still suffer discrimination and violence (Human Rights Watch 1999) and are behind other groups on many indicators of well-being (see Dreze and Sen 2002, who discuss this issue in several places and provide many references).

²⁰ Note that this need not be the birth order of the child within the household (for example, the oldest child in the household might not be attending school, in which case he/she would not be considered in the regression). The oldest child might also be married or attending an educational institution in a different region, and therefore be counted as part of a different household. Since the importance of birth order in the Indian context has been highlighted in some studies (see, for example, Monica Das Gupta 1987), we used the birth order of a child as a control variable and found that the results were essentially similar.

²¹ It is possible for two or more children living within the same household and attending school to be of the same age. In this case, we break the tie by assigning the child with the lower serial number (which uniquely identifies each member of a household in the ITUS) the lower birth order.

²² We do not present results for a probit estimation given that there are some serious problems with “fixed effects” estimation of probit models. The issue of whether sample weights should or should not be used in regressions is contested (Angus Deaton 1997: 63–72). We report here the results from unweighted regressions. The low R-squared values indicate much remains to be explained. However, we can reject the hypothesis that the model is invalid (i.e., an F-test).

²³ Almost all the data available on children in the ITUS is categorical (for example, boy or girl, and nature of relation to the household head). Hence the controls in the regression are all dummies. We performed a regression with a continuous variable (age), and the results were similar. As mentioned above, we also used birth order, which is a non-dummy variable.

²⁴ Motiram and Osberg (2007) investigate supply-side and demand-side factors that influence attendance and human capital accumulation of children in India. The quality measures that we use here are the same ones used in Motiram and Osberg (2007); namely, the percentage of primary schools that have a pupil-to-teacher ratio higher than 50 and the percentage of primary schools with no building or a kuccha building made of insubstantial material such as bamboo or grass (this variable, quality of schooling facilities, is separate for rural and urban areas).

²⁵ We have not presented these results here in the interests of space, but they are available upon request from us. Note that this procedure works because we are essentially testing whether sample selection bias exists. Fortunately, we do not find evidence for this. Otherwise, we cannot use this two-stage method for correcting sample selection bias. For a discussion of the issues involved, see Wooldridge (2002: 581–5).

²⁶ A better way to handle this concern would be to use an instrument for the gender of a child, but we are not aware of a variable that could serve this purpose.

²⁷ Although improved access to low-skill, low-wage jobs would increase the opportunity cost of girls' time, the positive possibility is that improvements in labor market opportunities for women in the upper part of the wage distribution may provide greater incentives for investment in human capital acquisition (Stephan Klasen and Francesca Lamanna 2009). The response of individual families will depend both on whether it is low- or top-end jobs that actually become more available and on the values and constraints of particular families (such as their gender norms, time preference, and need for income). Even if some families choose to invest more in their daughters' human capital, if others choose the immediate cash generated from their daughters' labor, the enforcement of child labor laws remains important for equality of opportunity.