US social policy is on an unsustainable path, primarily due to spending on the elderly, particularly health care. Yet technological and globalization shocks to the labor market may require increased government spending on working-age adults and their children.  To restore fiscal balance and protect the neediest, we need accurate information about the economic welfare of our citizens.

Until recently, poverty statistics suggested that programs for the elderly helped to nearly eliminate elderly poverty. However, the Census Bureau's new Supplemental Poverty Measure (SPM), which fixes widely-recognized biases in the “official” poverty measure (OPM), puts the elderly poverty rate far above the OPM: 15.9% (SPM) vs. 9.0% (OPM). The SPM elderly poverty rate exceeds that for non-elderly adults and approaches that for children (18.2%).  These results have fuelled calls to expand assistance to the elderly. Can it be true that, despite massive health and income transfers, the elderly remain among the most impoverished Americans?

According to Census estimates, the much higher SPM elderly poverty rate is entirely due to the subtraction of medical out-of-pocket (MOOP) expenditures from income:  together, all other SPM adjustments lowerelderly poverty. Moreover, neither the SPM nor OPM counts health benefits as resources, even though the SPM’s addition of “in-kind” benefits such as food assistance is uncontroversial. The SPM’s exclusion of MOOP and health benefits from income reflects a decision to base the SPM on non-medical needs and non-medical resources. 

We assess whether the SPM’s treatment of MOOP expenditures and health benefits results in an improved poverty measure. By “improved” we mean a measure that better discriminates between the “needy” and “non-needy”, better reflects differences in the need of different groups, and better reflects changes in need over time. 

We critique arguments for excluding MOOP and health insurance from income, particularly the arguments that MOOP is “non-discretionary” and that “health care deprivation” should be separated from “economic deprivation.” We analyze relationships among savings, medical care, insurance (beyond Medicare) and MOOP, illustrating how MOOP canbe discretionary, and review evidence that some MOOP is discretionary, particularly for long-term care.

We review four empirical validation studies that assess the impact of subtracting MOOP from resources or valuing health insurance, including new analyses conducted by some authors in response to our requests. There is little evidence that subtracting MOOP improves a poverty measure’s validity against a variety of indicators of hardship. On the other hand, assets clearly matter for the elderly’s experience of hardship. Since assets fund MOOP expenditures, deducting MOOP from income exacerbates problems created by the failure of both the OPM and SPM to consider assets economic resources.

We conclude that subtracting MOOP expenditures and excluding the value of health insurance results in an inferior measure of poverty for the elderly.  We recommend that consideration be given to using differentiated approaches to poverty measurement for households with and without elderly members. Finally we argue that it is paramount to create the best possible single “official” poverty measure to guide fiscal and social policy.