Appendix 3A

Alternative Approaches to Accounting for Medical Care in a Poverty Measure

The approach recommended in this report for incorporating medical care into the measurement of poverty builds on the Health-Inclusive Poverty Measure (Korenman and Remler, 2016; Korenman et al., 2019; Remler and Korenman, 2023b). Alternative approaches have been proposed, including the incorporation of medical out-of-pocket spending into Supplemental Poverty Measure (SPM) thresholds; multiple-index methods that treat health risk and needs as a distinct dimension of poverty; and a full-income approach. Table 3-1 provides an overview of the treatment of health and medical care in these three measures as well as in the Official Poverty Measure (OPM), the SPM, and the HIPM. Here, this issue is reviewed in greater detail.

Official Poverty Measure: The OPM does not explicitly treat health, medical care, or health insurance as needs. Government transfers of health insurance and employer payments for health insurance on an employee’s behalf are not treated as household resources; and medical out-of-pocket (MOOP) spending is not deducted from resources. Thus, the OPM mostly ignores medical care, and changes in the OPM over time reflect neither increases in households’ medical care cost burden nor any offsetting effects of public health insurance programs such as Medicare and Medicaid.

Supplemental Poverty Measure: SPM thresholds do not include a need for medical care, particularly care paid by insurance. Instead, they include an implicit need for care or insurance paid for out of pocket, by subtracting health insurance premiums and other MOOP expenses (e.g., copays, deductibles) from resources. Expansions of public health insurance coverage or increases in medical care costs are reflected in the SPM only as far as they affect household out-of-pocket spending. There are no adjustments for underspending by those who are not insured or underinsured. In contrast, since premiums are subtracted from resources while subsidies are not included in resources, those who receive subsidized benefits could appear to experience higher levels of poverty.

Health-Inclusive Poverty Measure: The HIPM proposed by Korenman and Remler (2016), described in detail in National Academies of Sciences, Engineering, and Medicine (2019) and Korenman et al. (2019), accounts for health insurance/medical care in calculating thresholds and resources. Specifically, a need for health insurance, as a proxy for medical care, is explicitly added to the SPM’s expenditure-based threshold, while the value of health insurance benefits received from the government or an employer (e.g., Medicare, Medicaid, employer-provided coverage) is added to resources and a capped amount of MOOP spending is subtracted from resources. The HIPM and the proposed approach for incorporating it into the Principal Poverty Measure are described in detail in Chapter 3.

Supplemental Poverty Measure with MOOP in Threshold (SPM-MIT): The SPM-MIT is conceptually similar to the SPM, as the name suggests. The main difference is that, instead of subtracting MOOP costs from household resources, a measure of the need for MOOP spending is added to the threshold, as originally proposed by Bavier (2000) and implemented experimentally by Short and Garner (2002) and Bavier (2006). One advantage of this approach relative to the SPM is that the threshold need for MOOP spending in the SPM-MIT is a predicted measure of a household’s expected need for medical spending, based on family size, age, and health insurance status (Garner et al., 2014)—in contrast to actual MOOP costs subtracted from resources in the SPM. This addresses a key concern about the SPM’s treatment of uninsured households, which may have significant unmet needs for medical care that would not be captured by subtracting actual MOOP expenditures from resources. However, like the SPM, the SPM-MIT is limited in its ability to capture the impact of any expansion of public health insurance programs on poverty; the SPM-MIT will capture such impacts only as far as they affect expected MOOP spending.

Two-Index Measure: In 1995, the National Academies panel recommending the SPM wanted to include medical care in the measure of poverty but recognized multiple validity problems in doing so. The panel suggested using a two-index measure, in which the second index would capture the medical care economic risk (MCER) to the population relative to the adequacy of their health insurance coverage to pay for their medical care needs. The Institute of Medicine and National Research Council (2012b) developed recommendations to estimate the proportion of the population at risk of incurring high MOOP expenses in relation to their resources as a second index to accompany the SPM, to better measure economic vulnerability. The SPM would serve as the core measure of poverty and include MOOP expenses as a deduction from resources, while MCER would serve as a second index that would capture the economic risk from having no or inadequate health insurance. The poverty status of an individual or family could then be determined by the core index or by a combination of the two indices. Using two indices would remove the issue of fungibility of health insurance as a resource in the core measure.

Technically, the MCER could be based on individual risk aggregated up to families. Individuals could be divided into a set of categories used for predictions of economic risk based on age group, sex, and broad measures of health. Such an approach would be useful to measure reductions in economic risk from medical expenditures due to changes in policy, such as the Affordable Care Act, Medicaid, and Medicare. However, this approach implicitly requires using the two indices together to measure poverty, which may not be politically viable; the approach does not address underutilization of care by the un- and underinsured; and it is difficult to measure accurately with current data sets.

Full-Income Poverty Measure: Burkhauser et al. (2020) define a Full-Income Poverty Measure (FPM) to evaluate the impact of the War on Poverty. The year 1963 is treated as the benchmark for this evaluation, which can be used to capture the impact of a wide range of antipoverty programs, including those that provide in-kind benefits (e.g., Medicare and Medicaid) and those that are administered through the tax code (e.g., the Earned Income Tax Credit). As Burkhauser et al. (2020) discuss, neither the OPM nor the SPM is suitable for this purpose, not only due to the failure to fully incorporate the value of in-kind transfers (the OPM excludes all in-kind transfers and the SPM excludes Medicare and Medicaid), but also due to changing thresholds over time, which deviate from the standard used in 1963.¹

In the FPM, resources include market income plus cash transfers, the market value of in-kind transfers (including Medicare and Medicaid), the market value of employer contributions to employees’ health insurance premiums, and tax credits—minus federal income and payroll taxes. Thus, in terms of its treatment of health and medical care, FPM resources are more inclusive than resources in the SPM, which reflect Medicare, Medicaid, and employer-provided insurance only as far as they reduce households’ MOOP spending.

On the threshold side, the FPM sets the threshold to achieve the official poverty rate in 1963 of 20 percent. This threshold is then updated annually using the Personal Consumption Expenditure price index. A need for health insurance or medical care is not explicitly included in the threshold; these goods are implicitly included in

___________________

¹ The specific reasons why the OPM and the SPM are not suitable for this purpose differ from each other, and interested readers are referred to Burkhauser et al. (2020, pp. 25-27).

the baseline threshold to the extent that they were being consumed in 1963. This approach embraces the idea that the correct yardstick for measuring the success of the War on Poverty is its effectiveness in reducing what would have been considered poverty in 1963, updated for rising average prices.

Over time, due to technological advances in medical care, an absolute poverty threshold that includes medical care becomes less meaningful. The 1963 medical care “good” held constant in the absolute threshold bundle becomes inexpensive to purchase (if it is even available). The health insurance benefits included in resources valued in today’s dollars buy care that is far more valuable than the now inexpensive but antiquated care of 1963 represented in the threshold. As a result, over time, a growing health benefit far exceeds the threshold medical care need and is assumed available to meet nonmedical needs. This distortion of the FPM is the result of an interaction between absoluteness of the medical care need in the threshold and the fact that health insurance benefits are not truly fungible resources; and, over decades, the distortion can grow large. The result can be seen in FPM trends between 1963 and 2017: the value of Medicare benefits for an eligible couple in 2017 exceeds the entirety of the couple’s FPM threshold. Obviously, while Medicare benefits are a highly valuable resource, they alone cannot meet all basic medical care needs, as implied by FPM analyses. Thus, the current panel does not view the FPM’s treatment of health insurance as a viable alternative to the SPM for ongoing poverty measurement.

Two remaining approaches to modifying the poverty measure are not included in this Appendix because they do not explicitly deal with health or health insurance. These are: (1) the SPM-absolute measure (Wimer et al., 2016) which anchors the SPM threshold to keep it absolute rather than allowing it to change with the 33rd percentile of food, clothing, shelter, and utilities (no specific treatment of medical care or insurance is considered that differs from the SPM); and (2) the Meyer and Sullivan (2012) estimate of consumption poverty with and without including medical spending. Meyer and Sullivan prefer to exclude MOOP spending, arguing that it likely represents substantial need or lack of good insurance (p. 142), and that medical spending might better be viewed as an investment similar to education, and thus should be excluded from a consumption-based measure. They also explore using a fungible value of public health insurance.