Is Health Insurance Affordable for the Uninsured?

Summary of working paper 9281
Featured in print Digest

Based on a plausible range of definitions and assumptions, health insurance is affordable for between one quarter and three quarters of adults who are not insured.

A common explanation for the fact that 15 percent of U.S. citizens do not have health coverage is that a significant proportion of the population cannot afford to pay for coverage. Yet the "affordability" of health insurance is a subject that is poorly understood and that has received relatively little attention from economists. In Is Health Insurance Affordable for the Uninsured? (NBER Working Paper No. 9281), economists Kate Bundorf and Mark Pauly develop a framework for understanding affordability and, based on a plausible range of definitions and assumptions, show that health insurance is affordable for between one quarter and three quarters of adults who are not insured.

Adults in low-income households are certainly more likely to be uninsured than people in higher income households, but not all uninsured adults are in families with low incomes. If low income is defined by the federal poverty line, then only 22 percent of adults aged 25-64 who are uninsured are in families with incomes below the poverty line while 30 percent are in families with incomes three times the poverty level. At the same time, some people purchase health insurance despite having what might be seen as an inadequate income, and this has important public policy implications. There may be a case for subsidizing health insurance for those who cannot afford it, or for people who have insurance but for whom paying the premiums causes hardship. But, for a large proportion of uninsured people, health insurance is a matter of choice. Policymakers can either mandate coverage - as is the practice in most countries - or not worry about people who choose to bear the risk.

Bundorf and Pauly use two broad approaches, analyzing a range of assumptions within each, to examine the question of affordability. The first approach relies on the commonly applied definitions of poverty to identify adequate levels of family income. The second approach relies on peer comparisons to ascertain what similar adults do. The researchers examine the implications of adopting the different approaches by applying them to data from the 2001 Current Population Survey, which provides information on insurance coverage. Their analysis focuses on adults 25-64 years of age without public health insurance coverage.

The authors first look at what people can "afford," based on whether household income is above or below the federal poverty line (or some multiple of the poverty line), adjusting reported income for differences between insured and uninsured adults attributable to employer premium payments for health insurance. They find that the insurance-adjusted poverty rate for adults aged 25-64 in 2000 was 10.5 percent; on that basis, health insurance is unaffordable for 10.5 percent of adults aged 25-64. For the whole sample, using the poverty line as a benchmark, 71 percent of the currently uninsured population could afford health insurance coverage. Increasing the definition of affordability to family income exceeding three times the poverty threshold, the proportion of "uninsured afforders" declines to 28 percent.

Bundorf and Pauly also present a number of estimates defining affordability thresholds according to the proportion of individuals with similar characteristics who purchase insurance. Using a definition of health insurance as affordable if the majority of people in similar circumstances purchase coverage, the authors find that health coverage was affordable to between 59 and 66 percent of the insured, depending on the characteristics used to define individuals as similar. Using the threshold that 80 percent of similar households purchase insurance, they find that around 25 percent of the uninsured could afford coverage based on peer comparisons.

Thus, the researchers conclude that the affordability of health insurance, measured in various ways, is not a particularly accurate predictor of whether a person will obtain coverage. It is certainly not the only explanation of observed patterns of insurance coverage. The broad picture that emerges from the authors' tests is that between 25 percent and 75 percent of people who do not purchase coverage could afford to do so. This provides a clearer framework for policy decisions and for prioritizing where public assistance is required.

Bundorf and Pauly's results apply only to those not covered by public programs. The researchers use a number of assumptions to control for various personal and family characteristics, but the wide range of the estimates also reflects unobservable differences in unmeasured income or wealth, and well as different preferences of the individuals and difference prices of health insurance that they face.

-- Andrew Balls