I have already written about some of the fallacies involving health care outcomes across countries and the alleged negative role of insurance companies and their pursuit of profit.
The third big area of fallacy and outright misinformation involves the measurement of health care expenditures in the US and as compared to other countries.
This trifecta of fallacies, pushed onto the American people by the alliance of proponents of health care change and the liberal media, have convinced most people that health care in the US is of low quality, that we pay too much for it, and that insurance companies are greatly to blame. Yet if you ask most people, they are actually quite satisfied with their own coverage. You know that the media have been successful when most people think their situation is fine but think that everyone else is really suffering.
As to the measurement of health care expenditures, there are myriad problems in just measuring how much one country (the US) spends, much less comparing that to other countries. The situation of cross-country comparison is compounded for health care relative to other goods and services because in many countries health care prices are controlled or the entire sector is nationalized. How can we use prices or wages in a country like the UK to measure how much of that nation's resources are being used in health care, when that sector is nationalized but other sectors are not?
But let's ignore that elephant in the corner for a little while and focus on some other interesting issues involving the US expenditures on health care.
If you look around for US expenditure data, you will quickly land at either the OECD site or the Research, Statistics, Data and Systems: National Health Expenditure Data site of the Centers for Medicare and Medicaid Services. The OECD has the cross-country data, but it takes a while to download the whole thing and learn how to use it. (Also, interestingly, as soon as you try to pull out data from more than one country at once, you are presented with a window saying: "Please be careful regarding cross-country comparisons. We advise you to consult the Sources and Methods concerning the data you have selected." I wonder how many people then check the box with "Do not display this window anymore."!)
The number for the US that is typically reported is that we spend $2.2 trillion, or 16.2% of our GDP, on health care. I have often heard that number, and always wondered if I should trust it. The answer of course is no, and I proved that to myself with just a bit of checking. Where should I start -- so many fish to catch, and so little time! The source of much of my discussion is from the Sources and Definitions document from the Health and Human Services site noted above. As we go through this, have in mind that to a great extent what that total expenditure number represents is a summation of what we spend as a nation when we buy health care: so when you go to the hospital and your insurance company (or employer if they are self-insured!) pays the bill, that gets counted as health care expenditure.
Now in all of this, I will note how important it is, when interpreting these data, to have a model of the health care industry in mind. A colleague of mine pointed this out, and it is really true. Without some theory to guide one on how, for instance, prices are set in the industry, the data by themselves really say very little.
1. Start with the fact that in that 16.2% number we have included $101 billion of capital expenditures on structures and equipment. The problem here is one of possible doublecounting, or at least inaccurate comparisons across countries based on how different countries account for capital spending and depreciation. Since our health care industry is either for-profit or not-for-profit (with essentially a break-even constraint) we have to assume that the prices of the health care services cover the capital costs. Think of a competitive equilibrium: prices must equal average total cost of providing the service, including the cost of capital. Since the government uses prices to measure how much we spend on health care, by adding capital expenditures again, there is some doublecounting. Interestingly, the OECD recognizes that capex is a different creature and recommends it be reported "below the line." This does not solve the problem in cross-country comparisons however. For a nationalized system like the UK, prices are unlikely to include a capital cost. Thus, if you don't add capex in a country like the UK, you don't even count cost of capital once. So leaving it out of the UK, and comparing that expenditure data to the US is still biasing the US as too high. This is not an insignificant amount of money.
So when a hospital buys an MRI machine, that total amount gets counted that year in expenditure data. Then I go to get an MRI, pay the hospital for that and in that price is an implicit payment for the machine, and that gets counted again!
(Note: In calculating GDP, capital goods do get added in, even though the cost of those will get reflected in prices. Depreciation gets subtracted when we move from GDP to net national product and national income. So in a sense, GDP has this same doublecounting problem, in that capital goods get counted once when they are bought and then again as they are reflected in prices of goods and services. The problem in health care is really then not so much with intracountry calculations but inter-country, when countries treat capital expenditures and prices differently.)
2. The data for the US includes the entire NIH budget and a good chunk of the NSF budget as well. For those of my colleagues in the life sciences who are getting all their research funded through grants, perhaps you will be surprised to learn that you are part of the problem. Of course, with "overhead" rates on grants from places like the NIH running close to 50%, there is also a lot of funding of general university overhead in the NIH budget. Cross country comparisons are rendered problematic when we realize that other countries fund their universities through direct payment rather than through the US' s reliance on "indirects."
3. Now here is a subtle one. But note again that the basic calculation for the US number is to add up how much we spend on health care. So when you go to the hospital and pay for a service, you pay for all costs incurred in the provision of that service -- wages, benefits, taxes, capital costs. Ah, to the extent that we have for-profit entities providing health care, those prices include corporate income taxes. Is that a cost of health care? Well, in the UK health services are exempt from the VAT, which is about 20% on the rest of UK GDP. Oops! Yet another not-insignificant incomparable.
4. And now another more subtle doublecounting. Again, those prices we pay for health care have to cover all the costs of the health care provider. Part of that cost is the cost of the health care benefits to the provider's own employees -- think of the staff of the hospital. So when a staff member of the hospital goes to the hospital herself, that gets counted once as a health care expenditure. But -- and I agree this is subtle -- that (expected) expenditure is already represented in the prices that are being paid by everyone else. Presto -- another very significant doublecounting.
Well, that should be enough to make anyone start wondering how much of GDP we really spend on health care, and if our spending rate is really that much higher than other countries. And again, the elephant in the room is that many other countries, if not all, have significant wage and price controls on the health sector, if not having that sector entirely nationalized. I have no doubt that the US could dramatically reduce its measured share of GDP spent on health care if we, say, reduced all doctors' fees and all drug prices by 25%. Is this the way we really want to go -- a system where our very imperfect political mechanism determines prices and profits for the people and companies that are protecting our health?