Massive transfer programs, especially the social security retirement program and the related supplemental security income system, have sharply reduced the poverty levels of aged Americans. In 1959 the poverty rate among persons 65+ was 57% greater than that of all persons in the U.S. (35.2% versus 22.4%). Thirty years later (1989) the rate was less than the population average (12.4% versus 14.2%). The incidence of poverty is not equal across the aging population, however. Citing a House Select Aging subcommittee report, a subcommittee member reported, "Women are 70 percent more likely to spend their retirement in poverty than men." (Columbus Dispatch, September 25, 1992) It is natural to ask how these women can be helped. At the same time, the huge expenditures required to secure the current reduction in poverty raises a second question of whether it is possible to achieve the same goal more cheaply.
To confront either of these policy issues, it is important to know the origins of poverty among retirement age women. Without an understanding of the processes that lead to poverty among the aged, policy planners must rely on increased direct cash transfers to the aged, perhaps through an expanded SSI program, as the only poverty tool. Is aged poverty primarily an extension of a life long condition or is it the result of negative wealth shocks later in life such as a divorce or a husband's disability or death? The first possibility is a basic redistribution question and is unlikely to be resolved outside a broader agreement on the appropriate distribution of income. The second is a social insurance problem and is potentially resolvable with changes in the design of the current social insurance system.
The National Longitudinal Survey of Mature Women provides a rich data set for exploring this issue. Offering a quarter of a century of detailed information on approximately 5000 female respondents 30 to 44 years of age in the first year (1967), the NLS panel provides a valuable opportunity to explore family income dynamics from midlife to the eve of retirement for the entire sample and into the retirement period for a substantial subset of the sample. The analysis focuses on the 1967-1989 period at which time the respondents were 52 to 66 years of age.