Application of experimental poverty measures to the aged

The U.S. Census Bureau recently released new, experimental measures of poverty based on a National Academy of Sciences (NAS) panel's recommendations. This article examines the effects of the experimental measures on poverty rates among persons aged 65 or older in order to help inform policy debate. Policymakers and analysts use poverty rates to measure the successes and failures of existing programs and to create and defend new policy initiatives. The Census Bureau computes the official rates of poverty using poverty thresholds and definitions of countable income that have changed little since the official poverty measure was adopted in 1965. Amid growing concerns about the adequacy of the official poverty measure, a NAS panel undertook a study of the concepts, methodology, and data needed to measure poverty. The panel concluded in its 1995 report that the current measure no longer provides an accurate picture of relative rates of poverty for different groups in the population or of changes in poverty over time. The panel recommended changes in establishing the poverty thresholds, defining family resources, and obtaining the required data. The Census Bureau report shows how estimated levels of poverty would differ from the official level as specific recommendations of the NAS panel are implemented individually and how estimated trends would differ when many recommendations are implemented simultaneously. It computes nonstandardized and standardized poverty rates. (The latter constrains the overall poverty rate under the experimental measures to match the official rate.) This article reports poverty rates that have not been standardized and provides considerably more detail than the Census report about the effects of the experimental measures on poverty among the aged. It examines the effects of changing the poverty thresholds and the items included or excluded from the definition of available resources. It also explores the effects of the experimental measures on persons aged 65 or older by age group, gender, race and ethnicity, and marital status. Results indicate that: Poverty rates in 1997 for persons aged 65 or older under the experimental NAS poverty measure are 17.3 percent, compared with 10.5 percent under the official poverty measure. This 65-percent increase is largely driven by the NAS-based measure's subtraction of medical out-of-pocket (MOOP) expenses from resources. Under the NAS-based measures, poverty rates increase for all major groups of older persons, and increase the most for groups for whom the incidence of official poverty is the lowest. The experimental NAS poverty measure shows narrower differences between genders, racial and ethnic groups, and among persons of different marital statuses than the official poverty measure. For example, white Hispanic women aged 65 or older have poverty rates that are 450 percent higher than those for white non-Hispanic men under the official poverty measure and 181 percent higher under the NAS measure. The NAS-based measure's subtraction of MOOP expenses from resources has a disproportionate effect on poverty rates among non-Hispanic whites and men as compared with other groups. However, changes in relative poverty between groups appear to be most influenced by the NAS midpoint equivalence scale. Because this scale decreases poverty rates for persons who live alone or with unrelated individuals and increases them for persons who live with others, poverty rates differ meaningfully under the NAS and official measures among demographic groups. This article highlights issues concerning the elements of the experimental NAS poverty measure that are particularly important to the measurement of poverty among the aged population. Results suggest that the research community's future efforts to refine, enhance, and build upon the NAS panel's recommendations will yield important insights about poverty among the older population.