Many Americans — perhaps most — believe that ours is the best health system in the world. That belief may rest on the fact that we spend more on health care than does any other developed nation in the Organization for Economic Co-operation and Development.

As a percentage of GDP and of per capita income, U.S. health spending is by far the highest in the world, even though the U.S. population is among the youngest among the developed nations. In 2017, total national health care spending in the United States amounted to about 17 percent of GDP, a claim projected to reach 20 percent by 2025. No other nation even comes close to ceding that large a slice of its economy to its health care sector.

The per capita cost of health care in this country has been rising inexorably over the past half century. Just how inexorable these cost increases are can be gauged by the ease with which drug companies can raise prices, even on long-established brand-name products and generics. Just how high U.S. health spending is, both in terms of total national health spending and per capita, can be seen from the international comparative data.

The OECD excludes certain items from “health spending” included in the spending data assembled and published by the Centers for Medicare and Medicaid Services (CMS) of the U.S. Department of Health and Human Services. Therefore, the OECD numbers are slightly lower than those published by the CMS, but the relative magnitude of spending among nations is indicative.

U.S. health spending per capita in 2017 amounted to an estimated $10,209 — about twice as much as is spent in most of the rest of the developed world. The major exception is GDP-rich Switzerland, whose per capita health spending is about three-quarters that of the U.S.

What Drives U.S. Health Spending to These High Levels?

Let’s explore the possibilities:

• Ability to pay, as measured by real GDP per capita
• The demographic structure of the U.S. population
• The high prices of specific health care services.
• The administrative burden loaded onto medical bills by the complexity of the health insurance system.

Ability to Pay

Our higher GDP per capita means we have greater ability to pay for goods and services, including health care, than poorer countries. Health care goods and services, however, differ from most in one important respect. Economists describe health care as a “superior good,” which means that spending on health care tends to rise disproportionately with income. If, for example, in a given year GDP rises by 5 percent, then typically health spending will rise by more than 5 percent, possibly with a lag of a year or two.

If we plot, for a given year, health spending per capita converted to U.S. dollars in terms of purchasing power against GDP per capita (ability to pay), we should not be surprised to find a positive correlation and an “income elasticity” (percentage change in health spending divided by percentage change in GDP) in excess of one.

The figure shows that GDP per capita does indeed drive health spending systematically. But income alone leaves much unexplained. Even after adjusting the health spending data for GDP per capita (roughly, “ability to pay”), U.S. spending levels are much higher (about $2,200 higher in 2015) than would be predicted by the graph above. So factors other than GDP per capita must be in play.

Demographic Structure

On average, per capita health spending first falls and then rises with age. It is so in every country, although the spending curve plotted against age varies a bit. Surprisingly, the age composition of the population is not a significant factor.

For the United States, the age-spending curve is indicative. Thus, one’s intuition forged by this chart suggests that per capita health spending in countries with older populations is apt to be higher than in countries with younger populations. Similarly, one would expect that the aging of the population within a country — for example, the graying of the baby boom generation in the United States — must be a powerful driver of increases in total per capita health spending averaged over all age groups.

But that is not the case, as can be inferred from the figure comparing age and health spending across countries. There is wide variance of spending per capita at any population-age level. The U.S. population is, on average, much younger than the populations of most countries in the OECD, yet we spend much more per capita on health care. In fact, although the United States has one of the youngest populations among developed nations, we have (as noted earlier) the world’s highest health spending per capita. Japan, in contrast, has the oldest population, but among the lowest health spending levels.

It is fun (though not really informative) to plot a “least squares” line on the data points in the figure. That line would slope downward, suggesting (spuriously) that the aging of a nation’s population lowers its per capita health spending. Just two data points would drive that result: the high per capita health spending in the United States, given its young population, and the low health spending per capita in Japan, given its elderly population.

Two-dimensional graphs implicitly reflect mathematical models with only one explanatory variable and one explained variable. But even in more elaborate cross-national regression models that control for many other variables thought to drive health spending, the estimated effect of aging on spending is statistically insignificant. It is a curious and counterintuitive fact, but one that should be noted.

The aging of the U.S. population is often fingered as a main driver of the annual growth in U.S. health spending. But is that really so?

Here, again, the answer is no. If one assumes a) that the age-specific, average utilization of health care per capita in dollar terms remains constant at some base-year level for the next three or four decades, b) that the total size of the population does not change, and c) that the only variable that changes is the age composition of the population, aging alone explains only about 0.5 percentage points of the usual annual growth of 4 to 6 percentage points in total health spending.

The twin figures below explain why. In the left-side version of the graph, the vertical axis goes only from 12 percent to 22 percent. That is the way the data are commonly presented in the context of analyses warning of the coming baby boom tsunami. By contrast, in the right-side graph with the vertical axis running from 0 percent to 100 percent, the much-feared baby boom wave appears as just a ripple. The aging of a population is just too gradual a process year over year to act as a major driver of the annual growth of U.S. health spending.

The High Price of U.S. Health Care

What, then, does drive higher income-adjusted health spending in the U.S.? With the exception of a few high-tech procedures, Americans actually consume less health care service in real terms (visits with physicians, hospital admissions and hospital days per admission, medications and so on) than do Europeans. For better or for worse — better for the vendors of health care and worse for consumers — prices for virtually every health care product and service in the United States tend to be twice as high as those for comparable products or services in other countries. The five figures on the opposite page make the point all too clear.

The High Administrative Overhead of U.S. Health Care.

No other developed country spends nearly as much on the administration of health care as the U.S., where, before the Affordable Care Act came along, administration absorbed 25 to 30 percent of total health spending. Most nations have relatively simple health insurance systems. Usually there is a heavily regulated social insurance scheme with uniform fee schedules and rules covering 90 to 95 percent of the population, along with a small private insurance market used by high-income households. The uniformity of fee schedules and rules creates ideal platforms for the smart application of health information technology that can lower administrative costs.

By comparison, U.S. health insurance is highly complex, with myriad insurance schemes that vary by the socioeconomic and demographic status of the insured, as well as by employer for families covered by employer-sponsored insurance. Consider how many insurance programs there are for near-poor elderly Americans. As shown in the figure on page 88, a system this complex makes incremental health reform challenging.

In particular, changing the rules in one part of the system can easily have an unintended effect on others. For example, establishing a new public health insurance program designed to fill the gap for working people whose employers do not offer them coverage might easily induce other employers who now do offer insurance to drop it and load their employees onto the new public program. This phenomenon can significantly raise the budget cost — sometimes by so much that it helps to kill the new program.

If one were to depict a so-called value chain for U.S. health care, it would look something like the figure on page 88. The clinical facet of the system is shown at the top of the chart. It is supported by a veritable armada of non-clinical actors, shown at the bottom. Each member of that administration believes itself to be adding “value” to the services delivered to the patient, who must pay for these camp followers’ often-handsome sustenance.

In an analysis of the growth of the U.S. health care workforce, Robert Kocher, of Venrock Capital, wrote in the Harvard Business Review:

My colleagues and I found that from 1990 to 2012, the number of workers in the U.S. health system grew by nearly 75 percent. Nearly 95 percent of this growth was in non-doctor workers. … So what are all these people doing?

Today, for every doctor, only 6 of the 16 non-doctor workers have clinical roles, including registered nurses, allied health professionals, aides, care coordinators and medical assistants.

Surprisingly, 10 of the 16 non-doctor workers are purely administrative and management staff, receptionists and information clerks, and office clerks.

The problem with all of the non-doctor labor is that most of it is not primarily associated with delivering better patient outcomes or lowering costs.

Private Insurers. According to a recent publication by America’s Health Insurance Plans, private health insurers on average take a haircut of about 17.8 percent off the insurance premiums paid by employers or individuals for “operating costs,” which means marketing and administration. Another 2.7 percent is profits. That haircut was as high as 45 percent pre-Obamacare for smaller insurers selling policies in the (non-group) market for individually purchased insurance. Under Obama­care the portion of the premium going to marketing, administration and profits was constrained to 20 percent for small insurers and to 15 percent for large insurers. Under the Republican Senate’s 2017 draft bill (The Better Care Reconciliation Act) states would have the freedom to relax these standards.

It’s worth noting that some analysts of administrative costs of Blue Cross Blue Shield plans and other private insurers report that the AHIP number is too high. They currently estimate the range to be between about 9 percent for the Blues and between 10 and 11 percent for other private insurers.

Public Insurance Programs. Government-run health insurance programs do not have marketing costs and profits, and they tend to have low administrative costs when calculated as a percentage of their spending for health care. On the other hand, with a never-ending stream of new, arcane regulations that require consultants for interpretation, they impose considerable administrative costs on doctors, hospitals and other providers of health care.

Moreover, the cut that private and public insurers take off the insurance premium and taxes is only part of the overall administrative overhead of our health system. And it probably is not the largest part.

Hospitals. Hospitals in other countries may have fewer than half a dozen billing clerks and coding consultants. By contrast, U.S. hospitals require hundreds or even thousands. Duke University’s health system, for example, with 957 beds, has 1,600 billing clerks!

Physicians. It is estimated that the typical American physician spends over $80,000 per year interacting with health insurers. That is nearly four times as much as is spent by colleagues in Canada and undoubtedly multiple times what physicians spend in Europe, Taiwan, Japan and Korea. In their paper, “U.S. Physician Practices versus Canadians,” Dante Morra of the University of Toronto and colleagues, wrote:

We estimated physician practices in Ontario spent $22,205 per physician per year interacting with Canada’s single-payer agency — just 27 percent of the $82,975 per physician per year spent in the United States. U.S. nursing staff, including medical assistants, spent 20.6 hours per physician per week interacting with health plans — nearly 10 times that of their Ontario counterparts.

Patients. None of the empirical estimates on administrative expense includes the value of the time American consumers devote to choosing health insurance products or, as patients, to process usually incomprehensible medical bills from the providers of health care or claims from health insurers. Patients who can afford it can hire billing consultants who front for them when querying medical bills.

Elisabeth Rosenthal, formerly of The New York Times and now at the Kaiser Family Foundation, explains that doctors and hospitals contract with armies of specialists to help them code their work in a way that extracts the maximum revenue from the rest of society (private and public insurers as well as patients). For their part, private insurers contract with equally adept coding experts to help them fend off “up-coding” by providers of care. There is nothing like this unproductive zero-sum game in the rest of the world.

The enormous administrative overhead in U.S. health care had been noted as early as the 1990s by the McKinsey Global Institute (MGI). The figure above is based on a very sophisticated study published by the MGI in 1996.

In 1990 Americans actually used $390 less in real health services per capita than did Germans. However, Americans paid $737 more per capita than Germans for identical health care products and services. Furthermore, Americans spent $360 more per capita than Germans on administrative overhead, and $259 more per capita than Germans for “other,” a category that may have included more administrative expense not specifically identified. In other words, almost all of the $390 in savings (the McKinsey authors called it higher productivity) squeezed out of American clinicians was absorbed in added administrative expense.

All told, Americans in 1990 spent $2,439 per capita on health care — $966 or 65 percent more per capita than Germans. Unfortunately, none of the higher U.S. spending represented more consumption of real health services. All of the difference (and more) between American and German health spending represented higher administrative costs, higher prices, and higher spending on “other” non-health care items.

It is anyone’s guess what added value American patients and insured consumers derive from the huge overhead they must finance. One can only hope that whatever that added value may be, it covers the enormous cost of that system. Alas, that is open to doubt. To quote Henry Aaron, the distinguished health economist at the Brookings Institution, in a commentary in the New England Journal of Medicine:

I look at the U.S. health care system and see an administrative monstrosity, a truly bizarre mélange of thousands of payers with payment systems that differ for no socially beneficial reason, as well as staggeringly complex public systems with mind-boggling administered prices and other rules expressing distinctions that can only be regarded as weird.

Our expensive drug distribution system. Spending on prescription drugs has become the fastest-growing component of total health spending in the United States. This is because Americans pay much higher prices for a given drug than do citizens of other developed countries. Furthermore, drug manufacturers in recent years have been able to raise prices substantially, even for long-established generic products where there is little or no ongoing R&D.

Recently the United Hospital Fund of New York undertook a study on spending on prescription drugs per insured member per month registered by a number of health insurance plans in New York State.

In their defense, drug manufacturers explain that the high cost of prescription drugs is partly the result of an elaborate and expensive drug distribution system.

Drugs that cost $17 to produce end up costing patients or purchasers of health insurance $100. Of a total $100 in consumer spending, health insurers pay pharmacy benefit managers $81, keeping $19 for themselves, of which $3 is profit. The rest goes for marketing and administration.

Pharmacy benefit managers are supposed to manage prescription drugs for health insurers. That task includes negotiating prices with drug manufacturers and dealing with retail pharmacies and patients. The PBMs pay pharmacies $76, keeping $5, of which $2 is profit. Of the $76 the retail pharmacies receive from the PBMs and co-pays from patients, they keep $15, of which $3 is profit. The retail pharmacies pay wholesalers roughly $60; they in turn pay manufacturers $58. Manufacturers spend $17 for manufacturing the drugs and $26 for marketing, administration and R&D; they keep $15 as profit, a handsome margin of 26 percent. In general, research-based drug manufacturers spend about twice as much on marketing and administration as on R&D.

Total profits alone booked by all of the agents in the value chain collectively amount to $23 of the $100 paid for drugs by consumers.

Although PBMs claim to be able to drive tough bargains with manufacturers and retail pharmacies over prescription drug prices, they seem to have only a modest capacity to do so. First, as can be inferred from data published by the International Federation of Health Plans, the prices Americans pay for drugs are the highest in the world. Second, the PBMs seem unable to resist the steep annual price increases for existing drugs that drug companies — even generic ones — routinely manage to impose on them. In early June 2017, for example, it was widely reported that Pfizer had raised prices on 91 existing drugs by an average of 20 percent since the beginning of 2017. There is a good chance that the PBMs will just accept such increases and pay them.

Insured patients typically are required to pay at the pharmacy a sum pegged to the price their PBM has negotiated with retail pharmacies. But the PBMs receive secret rebates from the drug manufacturers that the they claim are mostly passed on to the health insurers with whom they contract, who in turn claim to pass most of the secret rebates on to the employers with whom they contract. Given this secret rebate flow, one wonders what incentives the PBMs actually have to help keep drug prices low for consumers.

Probably nothing so clearly exposes the disregard for efficiency and costs with which Congress sometimes fashions health policy than the distribution of drugs for cancer and rheumatology treatments. These drugs are commonly administered on an ambulatory basis in the medical practices of oncologists and rheumatologists or in outpatient departments of hospitals.

Under the Medicare program (a major payer for these drugs) physicians and hospital outpatient departments are reimbursed for the drugs they infuse at the average sales price for the drugs reported by the drug industry to Medicare, plus a 6 percent markup over the ASP. With some of these drugs costing more than $100,000 a year, this 6 percent markup clearly provides a strong financial incentive for physicians to favor expensive drugs.

One would hope that in their clinical decisions most physicians would be impervious to this strong financial incentive. It was troubling, however, to behold the vehement opposition on the part of the American medical community, the pharmaceutical industry and members of Congress who front for the pharmaceutical industry to a proposal by the Innovation Center of the Centers for Medicare and Medicaid to experiment with alternative payment approaches.

The idea was to reduce the percentage markup on cancer and rheumatology drugs from 6 percent to 2.5 percent, and to increase the flat fee paid physicians and hospital outpatient departments for administering the drugs. That sensible idea had been recommended to CMS by the non-partisan Medicare Payment Advisory Commission (MedPAC).

In the face of this vehement opposition from powerful interest groups and from the members of Congress who carry their water, in late December 2016 CMS simply abandoned the idea of reforming this conflict-ridden drug reimbursement system. K Street lobbyists won out over plain common economic sense.

The role of Congress in driving up administrative expenses. I can think of no legislation ever to emerge from Congress that addressed the magnitude of this administrative overhead. It is as if Congress just does not care what health spending actually buys. On the contrary, every health reform emerging from Congress vastly complicates the system further and brings forth new fleets of non-clinical consultants who make a good living teaching clinicians and hospitals how to cope with the new onslaught. All of their income becomes the providers’ expense and thus ends up in the patient’s bill.

A classic illustration of this tendency is Congress’ introduction of Flexible Spending Accounts. They allow employed individuals to set aside a certain amount of money out of pretax income to cover out-of-pocket spending for health care in the coming year.

As is well known, or should be well known, any expenditure that comes out of pretax income or is deductible from taxable income is just another tax-financed subsidy in disguise. In this case, the larger the employee’s income and thus the higher their marginal tax rate, the larger that public subsidy.

Tax-preferred Health Savings Accounts are now all the rage. They, too, provide tax-financed public subsidies that mainly favor high-income households. Individuals can deposit money in these savings accounts out of pretax income to help defray out-of-pocket spending for health care. HSAs differ from FSAs in that unspent balances in one year carry over to the next, while unspent balances in FSAs are lost to the employee and revert to the employer.

Aside from being regressive, however, FSAs and HSAs have another dubious claim. These accounts have given rise to yet another profit-seeking new industry of non-clinical “health workers” who feed themselves at the health care trough, which patients must fill with their insurance premiums or out-of-pocket payments. Someone must hold the funds in the HSAs and make sure that the owners of FSAs and HSAs do not charge frivolous items not allowed under the legislation authorizing them.

Presumably, this new industry produces value for taxpayers who use these accounts to shift their tax burden onto the shoulders of other taxpayers, who are the losers from the policy. The industry’s revenues are part of GDP, as if it actually produced value for society as a whole. What that value would be, however, is not clear to this economist.

Nevertheless, Congress loves these non-productive redistributions of tax burden among households. Tax preferences — properly called “tax expenditures” by economists — are the vehicles for these dubious favors. It is easy to see why Congress loves tax preferences. As New York Times op-ed columnist David Brooks put it in one of his columns:

the best illustration of how the system [of tax preferences] works. Suppose the Pentagon wanted to buy a new fighter plane. But instead of writing a $10 billion check to the manufacturer, the government just issued a $10 billion “weapons supply tax credit.” The plane would still get made. The company would get its money through the tax credit. And politicians would get to brag that they had cut taxes and reduced the size of government!

So it goes. With tax deductibility of expenditures or other forms of tax preference, Congress can provide tax-financed subsidies to favored constituents that do not formally show up as government spending but can even be deceptively styled as tax reductions. Only economists and politicians appear to understand this dubious game.