Cross pressures on middle-aged adults: a broader view.

Abstract

A MID-LIFE wedged between obligations to •**• young children and parents would seem to be an obvious consequence of improved survivorship at the older ages. Despite the intuitive appeal of this argument, both the nature and prevalence of competing generational claims in mid-life remain open questions several decades after the notion first surfaced (Uhlenberg, 1993). Elsewhere in this issue, Rosenthal, Martin-Matthews, and Matthews (1996) use Canadian survey data to contribute new estimates of how many middle-aged adults balance potential filial duties with competing social roles. How common are three-generation families in which the dependencies of the youngest and oldest kin vie for the resources of the middle generation? At the most basic level this is a question of population structure, a fundamental concern of demographers. In the aggregate, the relative support burden on middle-aged persons (defined as broadly as 15-64 years of age or as narrowly as ages 45-49) is crudely measured by various dependency ratios. Estimates are regularly published for the total dependency ratio (the sum of youth and elderly support ratios) and, even recently, a sandwich generation ratio (U.S. Bureau of the Census, 1996). Because such calculations do not represent actual family linkages, these ratios represent per capita burden under the assumption that all middle-aged persons in a population contribute equal effort to support young and old, even those without children or living parents. More realistic estimates of the generational structure of families can be derived from simulation models. Multi-state life table models, for example, allow demographers to estimate the number of person-years a cohort survives with both old parents and children given a regime of nuptuality, fertility, and mortality risks. Watkins, Menken, and Bongaarts (1987) use such a macro-simulation model to show that the demographic conditions which prevailed in 1980 tipped the generational seesaw in favor of living longer with parents aged 65 and over than with children under age 18. On average, only five years were lived having both an older parent and a minor child, only slightly more than the overlap years under the 1900 schedule of rates. Microsimulation models, such as SOCSIM described by Himmel, Wachter, and McDaniel (1981) and Wolf's KINSIM (1988, 1994), generate even more refined representations of family trees in historical and projected time. Until recently demographic methods were required to estimate the simple prevalence of three-generation families within the population. But with the advent of the 1986-87 National Survey of Families and Households (NSFH), and its 1992-94 follow-up, data identifying kin both within and outside the respondent's household are available to describe the generational structure of mid-life families. Because family information is reported for spouses/partners if married, these data more fully describe the cross-pressures potentially buffeting mid-life families. Figure 1 displays the distribution of family structures by age for a mid-life reference population using data from NSFH 2. At ages 50-54, for example, about 5 percent have parents or in-laws but no children and about one in five have children but no parents or parents-in-law. By far, the most common family structure at ages 50—54 is the family containing at least three generations. Nearly 74 percent in this age group have one or more living parents (or in-laws) and one or more children (or stepchildren). The prevalence of four-generation families (parents, children, grandchildren, and the mid-life reference generation) is shown with a dotted line to indicate that this type of family is a special case of families with three or more generations. The prevalence of the four-generation family crests at ages 50-54, about 10 years after the peak in the prevalence of families with at least three generations. Unlike the Canadian data of Rosenthal, Martin-Matthews, and Matthews (1996), the estimates shown in Figure 1 summarize potential obligations to parents and in-laws and children and grandchildren. Several important features of contemporary family structure are obvious in these data. First, membership in a three-generation family containing older parents and children is modal for middle-aged adults up to age 60. Four-generation families also are fairly common for those in their forties and fifties. Second, with a mean age of childbearing hovering around 26 years since 1960, very little of the generational overlap occurs at a point when both the parents and offspring of middle-aged adults are likely to need care. Rather, the timing of parent care is more likely to coincide with the launching of young adult children.