Wealth and the affordability of health insurance

Between 2000 and 2005, the number of uninsured people in the USA has grown from 40 million to almost 47 million [1]. This has led to much debate over how many of these people can really afford insurance but refuse to purchase it. And for good reason. There exists a phenomenon that I call the ‘19–19% affordability puzzle’. That is, 19.3% of nonelderly adults in the USA (3.6 million) in 2002 with household income below the federal poverty line (people not expected to be able to afford insurance) did in fact buy insurance. By contrast, 19.7% of nonelderly adults (31 million) above the poverty line were uninsured [101].

How are we to assess this paradox? Many ‘unafforders’ do indeed buy insurance, and many ‘afforders’ do not buy insurance, if one arbitrarily takes the definition of affordability to be the federal poverty line. As Bundorf and Pauly show, approximately 75% of the uninsured should be able to afford insurance under this definition [2]. But, perhaps the poverty line does not capture affordability. Indeed, the original computation of the federal poverty line in the early 1960s was based only on the cost of food and did not include medical expenses. Even now that the poverty line is based on the Consumer Price Index (CPI), the growth rate of medical expenditures has far exceeded the growth in the CPI. For example, the nation’s hospital bill increased by 90% between 1997 and 2005, while the CPI only increased by 20% [102]. Thus, maybe the poverty line is too antiquated a measure to properly gauge the affordability of health insurance.

In fact, recent research has demonstrated that the high cost of medical care has indeed led to a very large financial burdens for families. In 2003, 24% of the nonelderly population living below the poverty line spent more than 20% of their after-tax disposable family income on healthcare. By contrast, only 1.6% of the nonelderly population living above 400% of the poverty line spent more than 20% of their disposable income on health [3]. This 1.6% burden for households with income over 400% of the poverty line does not seem so bad. So, suppose we use 400% of the poverty line as a generous definition of affordability? Unfortunately, this expanded definition of affordability will not explain the uninsured. In the 2002 Medical Expenditure Panel Survey, I found that 58% of uninsured nonelderly adults were actually above 400% of the poverty line, with an average household income of US$43,500. In comparison, the median household income for nonelderly adults in the USA in 2002 was only $42,300. Thus, most of the uninsured adults are in households that earn more than the median American household.

Can these people living above the median income really not afford insurance? Just how expensive is health insurance? The average annual family out-of-pocket premium in 2003 was $1644 for those with an offer of insurance from their employer (the total premium was $9249 after including the employer’s contribution). In fact, in 2005, 23% of private-sector establishments with offers were generous enough to offer at least one insurance plan requiring no contribution from the employee for family coverage [103]. But, for those families without an employer offer, the annual out-of-pocket premium was considerably more, costing $3650 for a nongroup (individual) policy [3]. Thus, maybe the uninsured people above 400% of the poverty line are simply people without an offer of insurance from their employer? In fact, I find that 79% of the uninsured above 400% of the poverty line do not have an offer from their employer. Thus, we see that a main source of the affordability problem is that people without an offer for insurance from their employer simply face a much higher price for insurance. This can best be remembered as the ‘85–85% insurance offer rule of thumb’. Overall, approximately 85% of the uninsured do not have an insurance offer from their employer and face the higher priced nongroup insurance market, while 85% of the insured do have an employer offer and face a lower premium.

So, what explains the behavior of the odd 15% groups of other people: the 15% of the uninsured who turned down an offer for group insurance and the 15% of the insured without offers who nevertheless bought nongroup insurance? I call this the ‘15–15% affordability puzzle’. That is, by moving from the poverty line definition of affordability to an employer-sponsored offer definition, we have reduced our initial 19–19% affordability puzzle down to a 15–15% affordability puzzle. Interestingly, income does not help to explain this 15–15% puzzle. In fact, they both have about the same income level: the 15% of the insured without offers but who bought nongroup insurance earned $60,606, while those 15% of the uninsured who turned down offers earned $61,618 in 2002.

Can we make any progress beyond this in explaining the affordability puzzle? In fact, we can. If we dig deeper, we find that affordability depends not just on disposable income, but also on assets and debt. People’s purchasing behavior is heavily influenced by their liquid savings, debt, stocks and bonds, individual retirement account savings, home and business ownership. Net wealth assets and net debt do not necessarily track one-to-one with income. For example, using new wealth and asset data from the 2002 Medical Expenditure Panel Survey, I found that 15% of the nonelderly adults in the lowest quartile of income (<$17,000) have a net wealth above the median net wealth [101]. Similarly, 10% of the top income quartile (>$67,000) are below the median net wealth. This variation between income and wealth might help to explain the affordability puzzles in health insurance.

For example, for nonelderly adults with income in the lowest 10th percentile of income (<$7000), enrollment in health insurance varies dramatically by wealth. For those in the lowest wealth quartile (<$1000), 53% are uninsured, as opposed to 31% in the highest wealth quartile (>$135,000). This pattern persists even in the top 90th percentile of income (>$101,000): 13% in the lowest wealth quartile are uninsured, and 5% in the highest wealth quartile are uninsured [101]. Thus, people do indeed seem to be purchasing health insurance based on net wealth, not just on income. This is in accordance with the consumption theory from economics, which says that people smooth out their purchases based on long-run permanent income and accumulated wealth rather than let their consumption fluctuate with their temporary income shocks.

In fact, the disparity in wealth between the insured and uninsured, even within the same income categories, is startling. Overall, the privately insured have $91,710 in net wealth, versus $9,490 for the uninsured. Within the income class of $15,000 or less, the insured have a new worth of $6400 and the uninsured have $2990 in wealth. For those with income in the range of $30,000–49,000, net wealth varies from $45,990 for the insured to $15,470 for the uninsured. For the high income (>$100,000) group, net wealth varies from $293,830 for the insured to $182,330 for the uninsured [101].

Let us now return to our affordability puzzle and look at wealth. In the 15–15% affordability puzzle, we first had a group of 15% of the insured who did not have an employer offer of group insurance, but who nevertheless bought nongroup insurance. Income does not explain this because their income was lower; only $48,800 compared with $70,650 for the insured with employer group cover. However, assets do explain their behavior. Their total net worth was much higher, at $289,170 compared with $200,964 for the insured-with-employer group insurance. Moreover, this was six-times higher that the net worth of the other people without offers who remained uninsured, $48,780, while their income was only twice as high as this group.

Next, let us look at the second group in the 15–15% puzzle: the 15% of the uninsured who turned down an offer for group insurance. Income does not explain this. Their income was $46,900 compared with $48,800 for those without an offer, but still bought a nongroup policy. However, their net worth was only $92,300, 68% lower compared with a net worth of $289,170 for those without an offer but who still bought a nongroup policy. Thus, we see that combining net worth with income explains a lot of behavior. How much does it really explain? Bernard, Banthin, and Encinosa estimate that for people without an employer-sponsored insurance offer, a model using only income would, for example, under-estimate health insurance enrollment for the low-income, high-wealth people by 26.7 percentage points [101]. However, a model that uses both income and net wealth would only underestimate enrollment by 4.4 percentage points [101]. This is a big difference. Especially, since these low-income, high-wealth people consist of the largest growing demographic group in our society: the baby boomers. This group, the near-elderly (aged 55–64 years), will double in size over the next decade. Moreover, they are often uninsured, since they very often lose their employer-sponsored insurance upon retirement while still being too young to join Medicare.

Does this insight on wealth shed any light on the policy debate over how to insure the uninsured? Yes, it provides a general insight that has often been overlooked. While most of the recent state reforms, such as universal coverage in Massachusetts and other similar proposals and programs in California, Vermont and Maine [1], have indeed been innovative, they often fail to distinguish between income and wealth when attempting to help poor adults outside of State Children's Health Insurance Program and Medicaid. As we saw above, using the poverty line as a gauge for health insurance affordability does not work. Family assets and debt must be taken into account. For example, the largest growing group of low income people are the baby boomers. However, since they have large asset holdings, such as stocks, bonds and real estate, they should be able to afford insurance even though their retirement income is very low. Thus, new nongroup policies targeted to help baby boomers obtain nongroup insurance should be means tested, perhaps with graduated tiers of premiums based on wealth and assets tiers, and not just on income.

In general, means and asset testing of premiums, deductibles and co-payments, based on assets as well as income, is an idea that the private sector must eventually explore. This will become especially more salient as the wealth disparity between the rich and poor continues to expand. It will no longer be sustainable for a $200,000 income worker to pay the same premium and co-payments as a $20,000 worker in the same firm. It is no longer just an issue of equity and solidarity, but an economic issue of affordability. The idea of means-testing premiums lost out in the early 1990s, when the private sector instead tried to base premiums to some degree on bad behavior, such as smoking and obesity. This has not worked; the obesity epidemic has continued to explode at an alarming rate. Moreover, many of the obese are poor. A more effective solution to treating obesity and other epidemics is to make healthcare more affordable to the poor by means- and asset-testing premiums and co-payments.

The private sector has realized this issue to some degree and has inched in this direction through the current movement called ‘valued-based healthcare’, where many efficiencies are gained by basing co-payments and premiums on the value of the treatment chosen by the patient. In this way, both patient and health plan share the cost-savings produced by the patient’s choices and good behavior. However, I would urge the private sector to go farther. Value-based healthcare will reach its full potential only when value-based programs are combined with income- or asset-based healthcare. Value-based premiums and copayments based on the value of the treatment could also be tiered according to the patient wealth and income. In that way, more patients will be willing to pay for insurance and participate in value-based healthcare. When patients remain uninsured, value-based insurance has no impact because most of the value is found in preventive care, which is the first level of care forgone by the uninsured. Thus, overall, as employers gradually move to a value-based healthcare system, I think it will be very beneficial for them to move concurrently to a means- and asset-tested form of value-based healthcare.

Disclosure

These views do not necessarily represent the views or policies of the Agency for Healthcare Research and Quality.

Financial & competing interests disclosure

The author has no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

No writing assistance was utilized in the production of this manuscript.