Widows Choose to Live Alone

Social Security accounts for nearly two-thirds of the rise in elderly widows living alone.

The share of elderly widows living alone rose from 18 percent in 1940 to 62 percent in 1990. But there is no reason for anxiety about the increasing unwillingness of younger generations to care for their parents, or the breakdown of the American family.

In Social Security, Economic Growth, and the Rise in Independence of Elderly Widows in the 20th Century (NBER Working Paper No. 6511), Kathleen McGarry and Robert Schoeni show that Social Security and other public transfers may "crowd out" the family, reducing the number of more traditional intergenerational arrangements. On the whole, people like their privacy: before 1935, many widows were too poor to live alone. But the improved economic status of the elderly, in particular the introduction and expansion of Social Security, has given widows the option of living independently.

During the 50 years preceding the Social Security Act of 1935, the living arrangements of elderly widows were virtually unchanged. Roughly 10 percent lived alone, 70 percent with adult children, and the rest in institutions or with other individuals. Intergenerational living arrangements began to change after the introduction and expansion of Social Security. The share of widows living with adult children declined from 59 percent in 1940 to 20 percent in 1990. Today Social Security is the most important source of income for the majority of widows. For widows not receiving Social Security income, or who have sufficiently low benefits, the Supplemental Security Income program provides a guaranteed source of income. Assistance of this sort was particularly important in the early years of the Social Security program when benefits were far from universal.

There also have been substantial changes in health, race, immigrant status, schooling, and fertility that could have had an effect on living arrangements. To isolate the effect of income, McGarry and Schoeni use evidence from the last six censuses, more than 50 years of observations. They incorporate a direct measure of Social Security income of the elderly over the entire period in examining the choice of living in an institution, living with others, as well as living with adult children or alone. They show that Social Security accounts for nearly two-thirds of the rise in elderly widows living alone. Changes in demographic factors account for a relatively small share of the change. Greater income translates into a higher probability of widows living alone throughout the period, but does not indicate a change in preferences over time.

Because Social Security and SSI may affect people with different incomes quite differently, McGarry and Schoeni consider both of them along with schooling (a proxy for lifetime wealth) as factors in the decision to live on one's own. They find substantial differences among groups of widows with different levels of schooling, both in terms of the level of independent living and in the importance of Social Security and SSI income on the decision to live independently. Widows with 13 or more years of schooling are over 5 percent less likely to live with children relative to those with 12 years of schooling. For widows with more than 8 years of schooling a $100 increase in monthly Social Security payments decreases the probability of living with children by nearly 7 percent. The effect is 1.15 percent smaller for those with just 0-4 years of schooling, who may have lower than average benefits or who may be less likely to have an earnings history that entitles them to benefits.

The most significant factor among sociodemographic changes is fertility. If the elderly do not live alone or in an institution, they most often live with children. Within the sample used, the number of children born decreased by 30 percent, from more than 4 in 1940 to around 3 in 1990. Additional children increase the probability that a widow will live with an adult child, lending support to the argument that a decline in fertility had been partly responsible for the changes in living arrangements. The authors find that an additional child increases the probability of living with a child by close to 3 percent.

-- Andrew Balls