Comment on charles F. Hohm’s “social security and fertility: An international perspective”

Abstract

Theoretical and empirical efforts to identify social and economic conditions that discourage high rates now include policy-relevant variables-programs that can be introduced by a government to reduce (Burch, 1975). Recent empirical work by Hohm (1975) suggests that government-administered old age social security programs will reduce fertility. The hypothesis that such programs lower is based on an assumption that parents regulate on the basis of the costs and utilities which children represent. Leibenstein (1957) first developed the view that children have a security utility to the extent that couples base childbearing decisions on the likelihood of having no other sources of support than their children during old age. As the state takes over responsibility for the aged, the security utility of children declines, bringing a lower birth rate. For example, Rich (1973, p. 43) asserts that alternative sources of social security can eliminate extreme economic dependence on children. As such systems become sufficiently stable for parents to rely on them, the importance of having a large family is likely to decline. For references expressing similar views, see Hohm (1975, pp. 630-631). While there may be a sound basis for expecting a social security effect on fertility, the importance of children for old age security in less developed countries may be overestimated. For example, family income in India (Mamdani, 1972) depends heavily upon the labor of family members, and a large family is an important resource for parents long before they reach retirement age. The idea of an old age security utility may be a modern invention, rather than an important force that regulates in traditional societies (Freedman, 1963). Hohm (1975) closely replicated the model of Friedlander and Silver (1967), while increasing the number of countries from 33 to 67. He concluded that . . the social security indices, on the whole, are at least as important as the control variables in predicting fertility (Hohm, 1975, p. 642). We first test the hypothesis that the negative relationships between 1965 total rates and social security program development found by Hohm (1975, Table 3) are simple artifacts caused by inadequate statistical controls. Actually, neither of the two coefficients for social security program measures using all 67 countries is significant at the .05 level (Hohm, 1975, Table 3).