Poverty

Perhaps of even greater concern is the question of whether the official poverty line sufficiently reflects the needs of older people, particularly in a society in which so many aspects of health care, long-term care, and other needs related to disabilities of aging depend on older people's own resources rather than public provision. Further, the poverty line for elderly individuals and for two-person, elderly-headed households is somewhat lower than the line for the nonelderly, although it is not clear that their economic needs are really less. For example, in 1999, for an older person living alone, the poverty line was $7,990, compared to $8,667 for a nonelderly person; for a two-person, elderly-headed family the poverty line was $10,075, compared to $11,214 for a nonelderly family. On the other hand, some have argued that the situation is even better for older people than the poverty figures show because the value of Medicare is not included in the poverty-line calculation.

The conventional poverty line has been widely acknowledged to be rather arbitrary in its construction. It dates from the 1960s, when it became important to planners of the War on Poverty to have a measure of a minimally adequate living level with which to define poverty and monitor progress toward the ambitious goal of eliminating it—an objective toward which there has only been real progress, ironically, with the older population, despite the War on Poverty's emphasis on children.

The basic idea underlying the original construction of the poverty definition, which was undertaken in the 1960s for the federal government by Mollie Orshansky, was straightforward to the point of being simplistic. Essentially, based on some studies that suggested that lower-income people spent about one-third of their income on food, Orshansky constructed food budgets for a minimally adequate diet, and multiplied these budget amounts by three. These original poverty line figures have been updated subsequently, based on Consumer Price Survey inflation adjusters. They have been criticized since their creation, both from the left (largely on the basis that a static measure of this kind does not incorporate increasing expectations for minimal adequacy in a society of increasing prosperity) and from the right (largely on the basis that they overestimate deprivation and have under-represented progress, since they do not incorporate the value of in-kind benefits such as Food Stamps, Medicaid, Medicare, and housing subsidies).

Those who believe that trends in the existing poverty measure present a more positive picture than they should have often argued that a relative income measure, such as a specified percentage of the median income, would be a more appropriate measure of trends in poverty than the existing poverty lines, because it would take account of the increasing cost of social participation at a basic level (automobiles, flush toilets, or air conditioning in Southern climates might be considered more of a necessity in 1999 than in 1965). Those who believe that the trends present an overly negative picture have often argued for valuing in-kind benefits in the poverty calculation. The argument here may be strongest for benefits like Food Stamps, which are near equivalents of cash. The argument for including health benefits seems more tenuous, as it does not appropriately address the greater health care needs of older persons, who even with Medicare spend more out-of-pocket on health care than do the non-elderly. Crediting each older individual with the average value of his or her Medicare benefit, as measured by average spending on behalf of all older people, attributes a large sum of "income" to each older person, though most are not receiving anything like that amount of services, since the distribution of Medicaid spending is highly skewed and a significant share of it takes place near the time of death. Such a calculation would distort cross-age comparisons by crediting the older age group with greater income because of their higher use of medical services. It would also distort time-trend analyses, making older people seem increasingly well-off as the cost of medical care services increases, thus reducing their income available for other purposes. If one attempted to attribute varying amounts of in-kind income to older individuals based on their actual or predicted Medicare expenditures, one would have the even more illogical result that older individuals get richer as they get sicker.

There have been a number of attempts to construct alternative poverty measures. A National Academy of Sciences (NAS) panel reviewed these issues and issued a report that recommended new ways to measure poverty. This report (see Citro and Michael, 1995) was critical both of the current measure's noninclusion of in-kind benefits and its failure to take account of the ways in which the cost of various goods, such as food, housing, and medical care, has changed relative to other goods since the early 1960s. Most recently, in response to such concerns, the Census Bureau has calculated poverty rates based on four alternative measures of poverty, one of which generally follows the NAS recommendations (Short, Garner, Johnson, and Doyle, 1999). This alternative measure adjusts poverty thresholds by geographical differences in the cost of living, counts noncash benefits as income, and subtracts from income some work-related, housing, and childcare expenses. The other three alternative measures include: the Different Child Care Method (DCM) assigns fixed amounts of child-care expenses to working families with children, based on the number and age of the children; the Difference Equivalence Scale (DES) uses a new method for adjusting for changes in expenses as family size increases; and the No Geographic Adjustment (NGA) measure, that, unlike the NAS measure, does not adjust thresholds for geographic differences in the cost of living. The alternative measures were standardized to the 1997 official poverty lines in such a fashion that each produced the same overall (all-ages) U.S. poverty rate as the official rate, and the measures were applied to income data for 1999. Interestingly, the 65+ poverty rate was considerably higher for each of the alternative measures than for the current official measure, while the child poverty rate was lower (there was little change for working-age persons). For example, while the 1999 poverty rate for persons 65 and over was 9.7 percent by the official measure, it was 13.3 percent (NAS), 13.0 percent (DCM), 13.7 percent (DES), and 13.6 percent (NGA) for the four alternative measures. By contrast, the poverty rate for persons under eighteen years of age was 16.9 percent under the official measure, but was 14.8 percent (NAS), 15.5 percent (DCM), 14.3 percent (DES), and 14.6 percent (NGA) for the four alternative measures. These results tend to support the argument that the official poverty lines underestimate the real prevalence of poverty among older people.

It is also important to note that the relatively low elderly poverty rate is not an indication that older people do not need Social Security; the rate is low only as a result of the presence of Social Security and would be far higher without it. For example, the Census Bureau found that without government transfers (mainly from Social Security) the 1999 poverty rate for persons over age sixty-five would have been 47 percent. Social Security has an important equalizing effect on income distribution—while there is evidence that inequality is higher among older people than at any younger age (Crystal and Shea, 1990), inequality would be far worse without Social Security. Social Security's tremendous success in reducing poverty in old age results in large part from its progressive income replacement design. It replaces a higher proportion of low incomes than of higher ones, while collecting the same contributions as a percentage of income (up to a ceiling)—a redistribution strategy that would be difficult to maintain to the same extent if part of Social Security benefits were replaced with individual private retirement accounts.