Unique features of the US health care system are unlikely to be a major cause of high US health spending (abbreviated version)

This is an abbreviated, bulleted, version of my recently published article in National Review on US health finance myths.   In the near future I will compile additional evidence for this line of argument and address probable objections across separate pages, so that I may tackle each at reasonable length, and combine them into a cohesive whole.  I will also be including links to data and code for those interested in doing their own analysis.  Please check back later.  In the mean time, the links contained in this post (below) are to outside websites and should go a long way towards addressing most of the more substantive questions and challenges.


There is a tremendous amount of misinformation surrounding US health care.  Much of this misinformation has become received wisdom, steered the debates over health care reform, and convinced many intelligent and otherwise reasonable people that we can save trillions of dollars annually and millions of lives if only we duplicate the health policy of other developed countries. Working from premises this misguided has the potential to be very costly.   Here are several myths that are in dire need of correction.

 Myth 1: High US spending must be explained by US health policy

Though it is true the US spends much more on health than the average developed country, it is also true that health expenditures are strongly associated with GDP and increase at a markedly faster rate than GDP.  In economic terms, health is a “superior good.” The US is an exceptionally rich country.  Accordingly, we should expect the US to spend an appreciably larger than average share of its income on health.

Using crude regressions on GDP, many have argued that US income levels still do not adequately explain US health expenditures. This is true, as far as such rudimentary analysis goes, but it is also misleading.  GDP is a measure of domestic production and it only predicts health expenditures to the extent that it accurately reflects actual material living conditions.   The relationship between these economic aggregates can vary quite substantially between countries for many reasons.  We routinely find the US much further ahead in comprehensive measures of household consumption and disposable income than it is in GDP.  The US stands apart from other developed countries in that it combines very high GDP with an above average disposable income share of GDP.


The relationship between these measures of material living conditions and national health expenditures (NHE) is consistently much stronger than the relationship between NHE and GDP.   Moreover, these alternative measures mediate the observed relationship between NHE and GDP.  In other words, once we account for either of them GDP has near zero marginal informational value, and they put the US quite comfortably within the normal range of variation.  There is little need to look for idiosyncrasies in the US health system to explain high health spending, as the spending is adequately explained by larger macroeconomic forces.


This does not imply that high NHE is necessarily socially optimal, or that we disregard the merits of proposed cost containment strategies, tax distortions, and the like.  Rather, it shows that health care finance is not a “solved problem” and that the potential to lower costs over the long run by mimicking the health policies of other countries is much more limited than proponents of reform would have you believe.

Myth 2: US prices are unreasonably high

The best available data, OECD health PPPs, suggest that the weighted average price of health care in the US is broadly consistent with what we would predict for a country of its wealth.  According to 2014 estimates, the price of health care in the US was just 10% more than the OECD average and very much on the income trend.


These patterns were observed across multiple rounds of study, and study indices were carefully constructed to enable reasonably reliable “apples-to-apples” comparisons across countries (especially making allowances for non-market transactions where conventional market prices are not meaningfully available).   In short, reliable price estimates show overall US health prices are not significantly higher than we should expect and that we spend much more because we truly consume much greater real volumes of care (2.1 times above the OECD mean in 2015).

By contrast, the figures typically cited by critics to suggest US health prices are obscenely high are quite unreliable and haphazardly selected.  They also often focus on labor intensive categories like surgical procedures, which the health PPP studies show to be much more closely linked to income levels, and fail to report prices for enough high-income countries.  Between the many sources of error most their data cannot be relied upon to draw reasonable conclusions.

Myth 3: Rising prices are the major cause of the relative rise in health spending

Although it is undoubtedly true that certain health prices have increased faster than incomes, standard estimates suggest that price increases explain between 0 and 22% of the observed increase in health spending between 1940 and 1990.   Likewise, since 1990 nominal GDP per capita has exceeded the average growth in the BEA’s Personal Consumption Expenditure (PCE) health price index, one of the most widely trusted domestic health price indices, which implies that prices explain none of the rapid growth in NHE over this period.  The evidence within and between countries is quite consistent: the relative growth in expenditures is overwhelmingly attributable to the consumption of a much greater volume of health care.  Additionally, a variety of associated indicators are quite consistent with this observation.

Myth 4: Physician cartels are to blame

Evidence indicates physician compensation accounts for less than 10% of NHE.  Physician pay has grown very slowly in the US over the past few decades, slower in relative terms than other OECD countries.  Physicians are paid multiples of the average wage in most developed countries and the US wage ratio is not especially high.  If the wage ratios US physicians enjoy is mostly a result of monopoly powers, instead of returns to skills, opportunity cost in medical training, lifestyle tradeoffs, and so on, the US is hardly particularly unusual in this regard and the price setting powers of central governments has seemingly done relatively little to curb physician pay.

Myth 5: Private insurance is to blame

According to CMS the net cost of health insurance, which is the difference between health premiums earned and benefits incurred, has averaged around 6% of NHE since 2003.   Included in this figure are profits and executive compensation (“bad” stuff in the eyes of some), but it is mostly vital expenditures such as contracting, claims administration, and fraud prevention, which ought to be included under any sane counterfactual.   This suggests that even if one regarded actual profit, executive salaries, and the like as completely worthless, the opportunity for cutting NHE by switching to an exclusively public system is greatly exaggerated.

To this point we can also look to per beneficiary expenditures: both Medicaid and Medicare cost substantially more per beneficiary than private plans (44% and 119% more respectively).  Admittedly these populations require more care due the fact they are older and sicker, but public plan expenditures are markedly higher than the OECD averages we would purportedly achieve if we scaled them out to the rest of the population.

Myth 6: Mediocre health outcomes must be explained by the health system

It is unlikely the differences in the US health system explain much, if any, of the observed shortfall in health outcomes for many different reasons.  First, conservative estimates suggest medical care explains less than 10% of the variance in health outcomes and factors such as genetics, behaviors, lifestyle, and social environment dominate.  Second, relative to other highly developed countries, the US is working against a large handicap in several crucial areas: high rates of obesity & diabetes, car accidents, homicides, drug use, historical rates of smoking, and more.  Third, there are some very large geographic differences within the US, which go a long way to explain the US shortfall.  These are very poorly explained by health insurance status and other indicators of health provision, but are very well associated with measured health behaviors and lifestyle.  Fourth, at least two quasi-experimental studies in the US have found that expanding access to health care has had no statistically significant effects on objective, measurable outcomes. Fifth, health access issues are apt to be much overstated as individual income does not predict markedly higher levels of health expenditure (somewhat the opposite) and even the year-round uninsured consume around half of what their fully insured counterparts do.  Sixth, international data strongly suggest that NHE is subject to rapidly diminishing returns with respect to objective outcomes like life expectancy and that most of the OECD has already passed some threshold beyond which the returns are consistent with zero.   Based on this data we would not predict a country with US NHE levels to achieve outcomes better than the median OECD country.  Indeed, incorporating known behavioral and lifestyle differences, we would expect it to be substantially worse.


There is much more that could be said on this topic, but suffice it to say that the weight of the evidence suggests proponents of single-payer are wildly overpromising.  This is not to suggest that the US health system is perfect, far from it, but that a root and branch overhaul is unlikely to deliver the promised goods.  Like all complex systems there is room for real improvement, however this is best pursued carefully and incrementally in a way that preserves the more dynamic elements of our system and rewards valuable medical innovation.  As in other highly developed countries, technological advancement is the main driver of health improvements in the US, and generates large spillover effects for the rest of the world.

2 thoughts on “Unique features of the US health care system are unlikely to be a major cause of high US health spending (abbreviated version)”

  1. Your data show that healthcare costs in the U.S. are basically in line with other OECD countries except that we consume healthcare at 2.1 times the average for OECD countries.
    I am trying to understand why we spend 18% of GDP on healthcare as opposed to maybe 12% or so for the next highest spending country. Is our extra consumption the cause of our huge overall cost?
    What I am most concerned about is our huge and growing national debt. (see http://itdoesnotaddup.com) The conventional wisdom is that the only way to get our debt under control is through entitlement reform, i.e. shrinking the cost of Medicare, Medicaid and probably also limiting the cost of the employer tax exemption for employee health insurance. Your data would seem to imply that this will be very hard to do.
    Is there any hope for solving our debt problem?

    1. I am trying to understand why we spend 18% of GDP on healthcare as opposed to maybe 12% or so for the next highest spending country.


      1) National health expenditures (NHE) are overwhelmingly determined by the average real incomes of households, not GDP (a measure of domestic production).

      2) The elasticity of NHE with respect to real household incomes is much greater than one, i.e., health spending grows in response to household income growth and it grows at a much faster rate the world over.

      3) The US is one of a small number of very high GDP countries.

      4) Conditional on GDP, real household incomes in the US are a bit above average whereas in the other very high GDP countries the situation is very much reversed.

      The problem with assessing health sending as a proportion of GDP and using it to draw comparisons between countries is that the real determinant of spending isn’t GDP and that the relevant denominator doesn’t relate consistently to GDP between countries (or even necessarily within the same countries in the long run).

      For example, over the past few decades, Norway’s GDP has hinged heavily on natural resource rents (oil, minerals, etc). They must continuously invest large shares of GDP into fixed capital to maintain production levels and even the free cash flows generated by these activities are very much temporary. Conditional on GDP, the true permanent income levels of Norwegian households are substantially below average, due to the capital intensiveness of these industries and the temporary nature of these flows. They need to save quite heavily if they don’t want their standard of living to drop precipitously in the coming decades (consumption smoothing). The state basically owns or controls most of these industries and takes the decisions out of the hands of individual households by siphoning off a large share of the free cash flows and investing in their sovereign wealth fund (most of which is invested outside of Norway), i.e., most of these cash flows don’t show up as household income that they can consume or invest.

      Luxembourg, on the other hand, is a small city-state with a large financial/banking sector and a large non-resident workforce. A large fraction of their GDP leaves their borders (on net) via their non-resident workforce and transfers of primary incomes (corporate transfers). Their apparent economic productivity in per capita terms is also heavily skewed as a result of this cross-border workforce (even my preferred indicators aren’t totally immune to this because the non-resident workforce obviously consumes a fair amount and buys stuff to bring home).

      Ireland’s GDP in recent years is also very misleading for similar reasons (primary and secondary transfers across borders). Foreign corporations have been using it for tax purposes, but most of this GDP isn’t truly available to domestic households….

      I’ve got to run, but I’ll try to address your other questions later.

Leave a Reply