On Measuring the Inequity of Financing Health Care in the United States and the Redistribution of Income Through Health Care Financing in Canada

As a society, we must care about not only how efficiently health care services are provided but also how equitably they are distributed. First, we want to know if health care services are consumed according to need or ability to pay. Different measures of equity in health care utilization exist depending on whether one considers specific services to be needs or wants. If the former, we want to see utilization independent of income; if the latter, we accept that high-income individuals utilize more than low-income individuals for the same level of need. However, even with that definition, we do not want utilization of the lowest-income individuals to fall below a certain threshold.

Second, we want to know if the burden of paying for these health care services is equitably distributed. A system in which low-income individuals use as much health care as those with high incomes but must sacrifice a larger fraction of their budget to do so could be considered inequitable. Again, this judgment depends on whether one considers health care services needs or wants. If the former, everyone should pay the same fraction of their income toward health care or there could even be some redistribution, meaning that the poor would pay a lower fraction than the rich to get their needed health care services. According to this principle, the income-related concentration of payments should not be smaller than the concentration of income. If the latter, where health care services are merely a want, the portion of an individual’s budget going to health care could increase with decreasing income without raising concerns about equity. Still, if individuals cannot afford food or housing because their expenditures on health care are so high, something called “catastrophic payment,” or if spending on health care makes them poor, the system can well be considered unequitable.

In an article for AJPH published in 2018, Christopher et al.¹ measured the (in)equity of financing health care services outside the public system in the United States. These health care services, representing approximately 60% of total expenditures on health care in the United States,² are financed by premiums paid for health insurance and so-called out-of-pocket payments (those not covered by any insurance policy). They use two measures: (1) the redistributive effect of spending, measured as the inequality of the distribution of income left once these costs have been incurred compared with the inequality of the distribution of income before these costs; and (2) the number of US residents falling into poverty as a result. Christopher et al. found that these payments outside the public scheme increase income inequality: paying privately for health care in the United States works as if it were a regressive tax, equivalent to moving 3% of average prepayment income from the bottom half of the US population ranked by income to those of the top half (authors’ calculations based on the formula for the Gini index).

As concerning as it is, this result is an underestimate, possibly a large one: if the higher-income individuals spend much more on health care that is not really needed (e.g., individual hospital rooms, cosmetic surgery), the amount of distribution from the less well-off to the better off will look much less dire than if measured on “needed” care only (e.g., cancer care).

That issue of underestimating the true amount of distribution favoring higher-income individuals of privately purchased health care seems intractable, as it requires a distinction between needed and wanted health care. To get around it, we can instead measure the redistributive effect of publicly financed health care (in Canada, hospital inpatient care, physician services and prescribed drugs for the elderly and those on welfare).

In a study conducted in Canada for the year 2011 (see also Corscadden and Wolfson⁴ for a slightly different version, using deciles instead of quintiles), Corscadden et al.³ simulated how much the average individual in each quintile of pretax income contributed to the amount of health care services purchased by Canada’s public single-payer system through income taxes and consumption taxes. Because of the conceptual difficulties in determining the incidence of corporate income taxes, they were ignored.

This analysis further posited that Canada’s federal and provincial governments use all the money in their total revenue indifferently to finance their various programs rather than allocating the money received from a given income quintile to a given program. We estimated the average cost of the health care services received by an individual of these same income quintiles, using survey and administrative data for inpatient care, physician services, and publicly covered prescription drugs (we used a survey with a very large sample size, greater than 100 000, to overcome the skewness of the distribution of health care costs). We then calculated the net “benefit” of each income quintile as the value of health care costs incurred in that quintile minus the amount contributed by the quintile. This assumes that the cost of health care reflects its value (and does so equally across income quintiles)—an assumption that can be disputed based on the numerous discrepancies between prices and (social) benefit in health care, but an assumption often made in equity analysis. Last, we calculated the effect of this net benefit on the inequality of the income distribution in Canada.

We expected to find a redistributive effect, from higher quintiles to lower ones: if the financing system is neutral and if, as is the case in any public system, how much one contributes has no bearing on how much one can use, we should see that the net benefit will be positive among the lower quintiles (who tend to be in poorer health and therefore use more) and negative among the higher ones, thus reducing income inequality. However, such an approach suffers from a bias of its own: it lumps together a measure of true redistribution across income levels and a measure of insurance or smoothing across the life course. This is because income and health both vary with age: after age 65 years, income tends to decrease at the same time that health care needs increase. A substantial portion of the “redistribution” effected by the public health care system can actually be considered deferred income, akin to public pensions. To take this into account, we also simulated lifetime profiles of net benefits for individuals, assuming that they remain in the same age-specific income quintile for their entire life. The effect remains substantial: a public scheme, collecting revenue based on ability to pay and distributing services based on need reduces the interquintile ratio (average income of the highest to lowest quintile) by 35%, from 6.6 to 4.3.

To obtain a full picture of the redistributive effects of health care in the United States or Canada, ideally we would need to measure the effect on the income distribution of both publicly and privately purchased health care. For the former, it is required to separate redistribution from smoothing and then to simulate lifetime benefits rather than cross-sectional ones. For the latter, it is required to separate the regressive impacts of deductibles and copayments for needed levels of care from the attenuating effect of higher-income individuals buying care that is not fully needed. Our view is that this task is intractable and that we would be better off measuring the number or proportion of households falling into poverty because of private costs of health care. This is what the study by Christopher et al.¹ does very well, showing that 9.3 million US residents fall into near poverty (and 4 million into extreme poverty) in a given year, because of catastrophic health care payment. Expanding private insurance, such as what the Affordable Care Act aimed to do, might help prevent these catastrophic payments (provided deductibles and copayments would be kept to a minimum), even though it cannot do much to reduce antiredistribution of private health care payments.