Affordability of the Health Expenditures of Insured Americans Before the Affordable Care Act

Abstract

Central to the Affordable Care Act is the notion of affordability and the role of health insurance in making otherwise unaffordable health care affordable. We used data from the 1996 to 2008 versions of the Medical Expenditure Panel Survey to estimate the portion of overall health care expenditures by insured respondents that would otherwise have been beyond their disposable incomes and assets. We found that about one third of insured expenditures would have been unaffordable, with a much higher percentage among publicly insured individuals. This result suggests that one of the main functions of insurance is to cover expenses that insured individuals would not otherwise be able to afford.

Central to the Affordable Care Act (ACA; Pub L No. 111–148) is the notion of affordability and the role of health insurance in making otherwise unaffordable health care affordable. Yet, to our knowledge, no reports in the health policy literature have estimated the extent to which insurance accomplishes this function. We used data from the 1996 to 2008 versions of the Medical Expenditure Panel Survey (MEPS) to estimate the portion of total health care expenditures by insured respondents that would have been beyond their disposable income and assets if they had been uninsured. We focused on the pre-ACA period because that period represents the political context in which the act was passed.

The MEPS data include information on demographic characteristics, medical care expenditures, health insurance coverage, incomes, and assets among a representative sample of US households. The information used in our analysis was derived from the household component of the MEPS, which is limited to members of the civilian, noninstitutionalized population who were present in the household during the entire survey period. We employed the restricted-use version of the MEPS to gain access to information on respondents’ assets.

METHODS

The health care expenditures of an insured MEPS respondent were deemed unaffordable if they exceeded the respondent’s income and assets after subtraction of the costs of other necessities. A number of alternative calculations were made.

On the expenditure side, base expenditures consist of the sum of direct payments to health care providers, including both out-of-pocket payments and payments made by private or public insurance programs. Self-reported expenditures tend to be underreported. We adjusted expenditures upward on the basis of a comparison of claims data from the MEPS and from MarketScan (representing the claims of employees of a sample of large firms).^1,2 Adjusted expenditures reflect this transformation.

With regard to income and assets, our base income variable is the income of the MEPS health insurance eligibility unit—that is, the household—on the assumption that any member of that unit would generally have access to all sources of income in the event of illness. This variable included income from wages, interest, businesses, farms, dividend payments, trusts, pensions and individual retirement accounts, Social Security and disability payments, veterans’ benefits, unemployment compensation, workers’ compensation, child support and alimony, payments from other public programs, and other regular cash income.

In the case of those insured through their employers, reported income does not include income paid in the form of insurance premiums. To capture the additional income those households would earn if they were uninsured, we added a premium payment to reported income on the assumption that, if such insured workers were uninsured, an employer would need to increase their take-home pay by an amount equal to the premium to be able to hire them. We used the average US premiums by household type reported in a Kaiser Family Foundation study to calculate this adjusted income variable.³

Some assets would also be available to pay for medical care in the event of illness, and the MEPS lists 2 asset variables: financial assets and net worth.⁴ Financial assets include bank accounts, money market funds, stocks and bonds, mutual funds, certificates of deposit, and retirement accounts. Net worth includes financial assets along with the net value of residences and other property, businesses, cars, and other savings (net of debts). Of these 2 variables, we relied on financial assets because they were more liquid.

With respect to the cost of other necessities, an amount equal to the federal poverty line was subtracted from income and assets to capture the resources available for health care spending after other necessities had been purchased. The federal poverty line, however, is an aggregate amount that cannot be disaggregated into amounts needed to cover the costs of different necessities at the poverty level, and so the portion representing poverty-level health spending could not be netted from it.⁵ Nevertheless, because poverty-level health care spending is relatively small, the federal poverty line still represents a reasonable measure of the resources available to spend on medical care.

Our empirical approach was to identify those respondents whose annual medical spending exceeded their household’s available resources. Because health care spending tends to be “lumpy” in the sense that a fraction of a procedure or course of treatment often produces much poorer health outcomes than none at all, we assumed that if any portions of expenditures are unaffordable, then all of the health care represented by these expenditures would be unaffordable. Thus, we summed the expenditures of respondents with any shortfall and determined the proportion of total insured expenditures represented by the expenditures of these respondents. Alternatively, we calculated the proportion using the net or unaffordable portion of expenditures among those households with a shortfall. Finally, we distinguished between those who were privately insured and publicly insured.

RESULTS

Table 1 shows the proportion of spending that would be unaffordable, were it not for insurance, under various definitions of unaffordability averaged for the even-numbered years (between 1996 and 2008) preceding the ACA. The table also shows the percentages of MEPS respondents (with unaffordable expenditures) who generated these estimates. Under definition 6, which we regard as perhaps the most defensible, about 36.5% of all insured health care spending would otherwise be unaffordable. Definition 7 includes only the shortfall of expenditures and is relatively similar at 34.0%. Those who are privately insured would be able to cover all but 20.5% of their spending. In the case of those who are publicly insured, 86.0% of their spending would be unaffordable were it not for insurance.

TABLE 1—

Mean Percentages of Health Care Expenditures That Would Be Unaffordable Without Health Insurance, by Unaffordability Definition, and Mean Percentages of Respondents With Unaffordable Expenditures: Medical Expenditure Panel Survey, United States, 1996–2008

Unaffordability Definition	Mean Overall % of Unaffordable Expendituresa	Mean % of Respondents With Unaffordable Expenditures
1. Adjusted income plus financial assets minus FPLb minus adjusted expenditures among individuals with private insurance (if less than 0)	20.5	3.0
2. Adjusted income plus financial assets minus adjusted expenditures (if less than 0)	29.9	4.5
3. Adjusted income minus adjusted expenditures (if less than 0)	33.7	4.9
4. Net adjusted income minus adjusted expenditures (if less than 0)	21.3	4.9
5. Base income minus base expenditures (if less than 0)	32.4	5.0
6. Adjusted income plus financial assets minus FPLb minus adjusted expenditures (if less than 0)	36.5	11.5
7. Net adjusted income plus financial assets minus FPLb minus adjusted expenditures (if less than 0)	34.0	11.5
8. Adjusted income plus financial assets minus FPLb minus adjusted expenditures among individuals with public insurance (if less than 0)	86.0	59.5

Note. FPL = federal poverty line.

^aAll percentages refer to even-numbered years between 1996 and 2008.

^bFor the various types of households by year.

DISCUSSION

Although the role of insurance as a mechanism for making unaffordable health care affordable seems obvious, until recently it has been eclipsed in policy discussions by the view that all of the additional care received by those who are insured is undesirable because it represents care that is valued less by patients than it costs to produce.⁶ Most public policies directed toward health insurance over the past 40 years or so—from the early imposition of coinsurance payments and the adoption of managed care in the 1980s to the more recent push for consumer-directed health care with large deductibles and health savings accounts—have focused on reducing this additional care. Although some additional care is clearly frivolous and inefficient in this sense, not all of it is, and this appears to be especially true of the serious, expensive health care procedures that become affordable only with insurance.

In our analysis, we focused on calculating the unaffordable portion of insured expenditures as a way of highlighting this function of insurance. In practice, this function appears to be diluted in our health care system by the availability of additional care for those who are actually uninsured. For example, Hadley and Holahan⁷ calculated that uninsured individuals taking part in the MEPS actually received 2.88 times what they spent out of pocket in health care services through nonspecific charity care or government programs.

Our study was also limited in that we did not identify all of the characteristics of insured respondents for whom this function of insurance is relatively more important. We did show that this function is particularly important for those who are insured through public programs such as Medicaid. Because the ACA focuses on expanding insurance to a similar demographic, it is likely that this function will be disproportionately achieved by the ACA. Increasing the affordability of health care is, after all, the ACA’s main intent.