Covariates of Household Extension in Rural India: Change Over Time

Abstract

The prevalence of extended households is still high in agrarian societies such as India relative to developed countries. The economic advantages to living in an extended household are many. First, an extended family may be better able than a nuclear family to support the aged, the ill, the widowed, and the disabled, since the cost is shared by a larger number of economically active members. Second, there are economies of scale associated with extended family living because resources and expenditures are shared. And third, work and responsibilities are shared, thus reducing the opportunity cost of childbearing and childrearing (Anderson & Allen, 1984; Cain, 1984; Ermisch, 1988; Sahai, 1973).These advantages would seem particularly attractive to households living in the Semi-Arid Tropical (SAT) regions characterized by subsistence agriculture, a poor resource base, and extremely volatile weather conditions. With lack of any strong institutional support, families may have to partly rely on their extended kin networks to help them cope with shortfalls in their incomes resulting from the risky environment and to ensure sustenance during old age and widowhood.This notion has been previously explored by Rosenzweig (1988a, 1988b). Using panel data for 10 years collected by the International Crops Research Institute for the Semi-Arid Tropics (ICRISAT), Rosenzweig found that households with high farm profit variability had greater intergenerational extension and relied on transfers (gifts, remittances, and receipts) from extended kin networks (Rosenzweig, 1988b).Using the same data set for 9 years on 201 households, Rosenzweig found that wealthier households were better able to self-insure and therefore rely less on transfer income. The landless, however, relied more heavily on credit than on transfer income (Rosenzweig, 1988a), indicating that perhaps they lacked the extended kin network for support.Despite the advantages of living in an extended household, individuals do not live in only one type of household throughout their lives. Their membership in nuclear or extended households is likely to be determined partly by their life cycle stage. Walker and Ryan (1990) reported that the prevalence of extended households in the ICRISAT villages has increased steadily over the panel period. They have attributed this increase to the aging of the panel households, indicating that household extension is primarily determined by the life cycle. Although life cycle is an important determinant, there are several other factors that could affect household structure, such as landownership (Barkat-e-Khuda, 1985; Caldwell, Reddy, & Caldwell, 1984; Nimkoff & Middleton, 1960), marital status (Anderson & Allen, 1984; Angel & Tienda, 1982), labor force participation of women (Pasternak, Ember, & Ember, 1976; Wong & Levine, 1992), and education (Conklin, 1988), and these factors need to be considered. Furthermore, prevalence of household extension could vary depending on changes in the characteristics of individuals over time, or on changes in their propensity to live in extended households, or both.The ICRISAT villages, over the panel period, experienced several changes at the village level that could also have affected household extension. For example, the distribution of landownership became more equitable in all villages, with small or medium farm households gaining more land through purchases or other means, and large farm households losing land. Groundwater irrigation increased in importance and in use, although cropping intensity did not keep pace. Seasonal and temporary migration by one or more members is considerable and has been increasing, especially due to the public works programs in some villages (Walker & Ryan, 1990).Our study seeks to examine change over time in household extension and in its determinants, taking into account the above-mentioned factors, using panel data for 9 years from six ICRISAT villages. …