Life-cycle analysis of household composition and family consumption behavior

Abstract

The purpose of this study is to assess the demographic impact of household composition on the consumption behavior of Nepali households over the family life-cycle, employing consumer demand theory. Household size and household composition are distinguished, and particular attention is focused on changing household composition and life-cycle consumption. Other factors examined are future household composition, demographic scaling, effects of different lifetime discount rates, and bequest motives. The life-cycle model of consumer demand theory originally advanced by Modigliani and Brumberg (1954) is employed and extended to investigate how changing household size and composition over the family life-cycle affects the consumption behavior of selected households. A sample of 614 households from Kathmandu Valley was taken during the winter of 1987-88 to represent the demographic structure and income-expenditure patterns of Nepali households. This study shows the life-cycle consumption of Kathmandu households in a changing household composition pattern. HOMES, A Household Model for Economic and Social Studies, is applied to project a family of curves of future household membership. This information is used to estimate the life-cycle consumption function that explicitly incorporates expected lifetime household membership. Equivalence scales are estimated to measure the effect of household age-sex composition on consumption. The findings confirm the importance of demographic variables in household consumption. The equivalent adult units that provide an alternative measure of the net effect on consumption of a specific household member are estimated. The estimated equivalent adult units for children, teenagers, and adult females are respectively 0.94, 1.46, and 0.92, assuming that the equivalent adult unit of all adult male is one. These large values of the equivalent units suggest that children have a stronger effect in Nepal than typically found in the literature. The demographic variables are jointly, but not individually, significant. This study is limited by a small sample size. Therefore, no firm conclusions could be drawn. Two factors may account for the teenage coefficients. First, teenagers do not contribute to the household income. Second, education in Kathmandu, particularly up to high school, is very expensive. Heavy outlays for tuition, fees, and school supplies wake teenagers more expensive to support than other members. Yet, the value of 1.46 cannot be distinguished from unity. Testing our model's general specification to examine the validity of the model and variables therein suggests that the relationship between demographic variables and consumption is not as simple as specified by the model. The equivalent adult unit for teenagers is also significantly different from one. This model also rejected the hypothesis that the bequest parameter is equal to zero. This confirms that the bequest motive is important and should be incorporated in the life-cycle model. The model developed here provides a better way for assessing the impact of current demographic characteristics on current needs relative to lifetime needs. The effect of demographic variables is not only through the existing age-sex composition of the household but also through the expected lifetime composition. The findings suggest that household composition has an important role in determining household consumption ratio, which in turn affects saving and economic development.