November 20, 2003

Government and Health

Posted by Brian

For amusement I was traipsing through the OECD health stats for various countries, and I was stunned by one of the things that springs out of the data - health care systems that are government run or funded tend to be cheaper despite being just as effective in every respect, and more effective in some respects. I'm sure someone somewhere has analysed the data properly, but even a crude analysis suggests the empirical case for having a government run or funded health care system is quite strong.



Here are the numbers, I'll explain them all below. (Apologies in advance if the table crashes your computer.)

Country Pub $ % FL ML IM PP
Australia 69.5 2224 8.7 81.8 76.2 5.7 2.5
Austria 69.3 2006 7.8 81 75.1 4.4 3
Belgium 72.2 2114 8.5 80.8 74.4 4.9 3.8
Canada 70.4 2433 9.1 81.7 76.3 5.3 2.1
Czech Republic 91.5 969 7.1 78.2 71.4 4.6 3.1
Denmark 82.2 2344 8.5 79 74.2 4.2 3.4
Finland 75.3 1608 6.9 81 73.8 3.6 3.1
France 76 2211 9.3 82.5 75 4.3 3.3
Germany 74.8 2615 10.6 80.7 74.7 4.5 3.2
Greece 53.4 1516 9.6 80.6 75.5 6.2 4.4
Hungary 78.1 771 6.8 75.2 66.4 8.4 3.1
Iceland 84 2559 9.5 81.5 77.7 2.4 3.4
Ireland 72.8 1623 6.2 79.1 73.9 5.9 2.3
Italy 72 1883 7.8 82.3 75.6 5.1 4.2
Japan 78.1 1852 7.5 84 77.1 3.4 1.9
Korea 43.1 762 5.6 79.2 71.7 6.2 1.3
Luxembourg 87.9 2685 6.1 81.1 74.6 4.6 2.5
Mexico 49.4 457 5.5 76.1 71.2 24.3 1.4
Netherlands 63.3 2310 8.7 80.5 75.3 5.2 3.1
New Zealand 77.5 1527 7.9 80.8 75.7 5.8 2.2
Norway 85.2 2550 8.5 81.1 75.6 3.9 2.8
Poland 71.1 558 6.2 77.2 68.2 8.9 2.3
Portugal 67.6 1469 8.7 79.2 72.2 5.6 3.1
Slovak Republic 89.6 666 5.8 77.2 69 8.3 3.6
Spain 72.1 1426 7.5 82.1 75.1 4.5 3
Sweden 85.7 2053 8.4 81.9 77.1 3.4 2.9
Switzerland 55.3 3080 10.7 82.5 76.8 4.6 3.4
Turkey 71.9 301 4.8 70.2 65.6 40.3 1.2
United Kingdom 80.5 1704 7.2 79.8 75 5.8 2
United States 44.2 4287 13 79.4 73.9 7.1 2.7
Correlations -0.09 -0.25 0.06 0.02 -0.22 0.19
w/o Mex & Tur -0.23 -0.40 -0.06 -0.05 -0.28 0.09
$ Correlations 0.84 0.61 0.69 -0.50 0.31
w/o Mex & Tur 0.80 0.50 0.63 -0.42 0.10

Here's what the columns mean:

Pub - Public Expenditure on Health as a Percentage of Total Expenditure on Health
$ - Total Expenditure on Health in US$ per capita (Update 20/11: These are PPP adjusted. Sorry I should have made that clearer earlier.)
% - Total Expenditure on Health as a % of GDP
FL - Female Life Expectancy at Birth in Years
ML - Male Life Expectancy at Birth in Years
IM - Infant Mortality - Deaths per 1000 Live Births
PP - Practicing Physicians per 1000 population

The figures in italics are from 1998, the others are from 1999, the last year for which complete statistics are available.

The lines in bold are the correlation coefficients between the first column (% of the health care system that is government funded) and the others. And it suggests there is a positive correlation between government involvement and (a) life expectancy and (b) number of physicians, and a negative correlation between government involvement and (a) health care costs and (b) infant mortality. The correlations with life expectancy are obviously very weak. Turkey and Mexico are outliers, so I took them out to see what difference it made. And the weak correlation with higher life expectancy goes away, but the other results are unchanged.

The last two lines are the same correlations, but now with absolute expenditure rather than government involvement as the baseline. The results are pretty striking. Throwing money at health care really produces positive results. And that holds up even if you remove the two outliers.

I'm sure someone has a more convincing version of this, but I think the prima facie case that increasing government involvement in the health care system, because it produces budgetary savings without harming the community's health, is pretty strong. But I'm sure plenty of people will have objections. Here are my answers to the most obvious ones.

Haven't other people looked at this in more depth? Why are you bothering with this?
I'm a blogger - it's my business to write about things I'm uninformed about. Besides, these numbers are pretty striking. I had always believed that government funded health systems are better in many respects, but I never thought there would be such a simple argument to that effect.

Isn't correlation a really crude measure to use here?
Sure. If someone has a less crude measure showing government involvement is financially or medically bad, it could easily overturn this. But it's certainly a pointer, and it's pretty unambiguous which direction it points in. (Besides, I'm a philosopher, we're not quantitatively trained!)

Aren't there good theoretical arguments showing that government is bad for you?
Well, there are arguments, but it's questionable how good they are. In any case, given a choice between theory and data I'll take the data 11 times out of 10. In Australia for a while the term 'economic rationalist' was being used for free-market true believers. I always thought there was a market opening for 'economic empiricist', because that's pretty much what I am.

Could there be some other explanation of the correlations?
Obviously there could be. It could just be the climate that explains why infant mortality in Iceland is so low, for instance. But note that the negative correlation between government funding and absolute funding removes the most obvious external explanation. If there was a positive correlation here then of course you'd expect there to be a positive correlation with good things like doctors and live babies. That the correlation remains despite the big government countries spending less money is remarkable.

Have you just cherry picked the data?
Perhaps unintentionally, but these were the only numbers I looked at. The only manipulation I did was to check the numbers without the two outliers, which actually weakened my case a little. (Not that correlations of the order of +/-0.06 mean much at all anyway.)

Does this support single-payer or single-provider or something else?
It's neutral. The data only measure how many government dollars go in, not whether they go to private or public institutions.

I know this isn't conclusive, but next time someone says government involvement in health care is bound to lead to budget blowouts, enormous queues, death, plague, or any other horseman of the apocalypse, I'll ask them to find some data to show that before I believe them.

Thanks to a commentator on Matthew Yglesias's blog for the link to the OECD site.

Posted on November 20, 2003 02:18 AM UTC
Comments

According to these figures, per capita public expenditure on health in the U.S. would be 44.2% of $4287, or $1895. That would put it sixth on the list, after Luxembourg ($2360), Norway ($2173), Iceland ($2150), Germany ($1956) and Denmark ($1927). The U.S. already spends more public money on health than France, England, Australia, Canada or Japan, among others. It could achieve both a total per capita expenditure and a public/private balance comparable to the countries just listed simply by cutting private expenditures by about $1760 per capita, with no change at all in the level of public spending. Do you suppose that would improve outcomes?

Posted by Vinteuil · November 20, 2003 03:15 AM

Well, I think just cutting private spending and not doing anything else would be a bit of a disaster. But I do think rearranging the health system to involve a bigger role for central government (e.g. Australian style Medicare, if not British style NHS) could have just that effect.

Put another way: Good God! Nationalising health insurance might mean both individuals and government save money, while health care standards get better! I should be shouting this from the rooftops. I mean, who knew the US government already spends more per capita on health care than the UK government?

Posted by Brian Weatherson · November 20, 2003 03:24 AM

It was enscribed:
…could achieve both a total per capita expenditure and a public/private balance comparable to the countries just listed simply by cutting private expenditures by about $1760 per capita, with no change at all in the level of public spending. Do you suppose that would improve outcomes?

No, because we have a perverse system. One of our perversities is that we spend far too much on administrative overhead at various levels; this is a lot of the difference. So, spending less money on the same system will not help. On of the main points of a government-run (or heavily influenced) setup is the reduction in overhead it makes possible.
What the numbers do show is that there is tremendous scope for improvement. Saying how that should be achieved needs to be done by someone who knows more than I. And we all know that the politics are probably impossible. But the gains are there to be made; at least, others have been able to make them. What do they have the we don’t?

Posted by Jonathan Goldberg · November 20, 2003 03:40 AM

You always have to be careful with these sorts of figures. In the UK, there’s a tax line item for the NHS. It is therefore often assumed that the NHS budget is made up only of that “contribution”. But of course the NHS also draws substantial funds from general revenues. Often, though, government statistics will list the NHS budget as merely comprising the revenues from that single line item.

The U.S. government plays similar games with many agencies, including the Postal Service. In years where the Postal Service makes a profit, it is moved “on budget”. In years where it makes a loss, it is moved “off budget”.

While I have no illusions about the cost of the labyrinthine U.S. health care system, whose insane insistence on tying coverage to employers still perplexes me, I find it hard to maintain a straight face when people blithely assert that everyone would be better off for less money were health care to be nationalized. Maybe at Hogwarts.

Posted by ASG · November 20, 2003 04:46 AM

Even if the OECD has been duped into understating the amount the British gov’t spends on health care (which I somewhat doubt), the data still shows that the US gov’t expenditures on health are within 10% of the total expenditures on health in the UK (on a per capita basis). But if you rely on just what the government provides in the US for your health, you’ll be much worse off than you would be given the available options.

I think it’s only at Hogwarts that the private sector could do as well as the gov’t at efficiently providing a health care system.

Posted by Brian Weatherson · November 20, 2003 05:02 AM

Brian, Jonathan:

What is to prevent the U.S. public sector from rearranging the financial resources it has already got? Given that those resources are already fully comparable to those of the best socialized systems in the world, the ratio of public to private expenditure seems like a red herring.

Incidentally, the particular outcomes listed in the chart are hardly the be-all and end-all. Keep in mind that a huge percentage of U.S. spending is devoted to end of life care that has little effect on life expectancy. The government could probably improve U.S. numbers by rationing its share of such care in favor of preventive measures, but what are the chances that the senior citizens lobby would sit still for that?

Posted by Vinteuil · November 20, 2003 05:12 AM

The usual conservative response to nasty data like this is to insist the US “free market health care system” is subsidizing the rest of the world’s socialist systems. No actual evidence is provided, but there you go.

Oh yeah, and hand waving about immigration. See: janegalt.net, chiefly.

Posted by Jason McCullough · November 20, 2003 06:00 AM

Since I don’t have any data whatsoever to hand, I’d better be careful with my assertions. But I think I’m right in saying that health care expenditure has hardly any impact on aggregate health outcomes (so much of it goes on people who are about the die anyway). Even expenditure on public health doesn’t affect things much. So there’s a bit of a problem in making assertions about the relative efficiency of health care systems where there’s little reason to suppose that those systems are crucial in achieving the outcomes their efficiency is being assessed by.

Posted by Chris Bertram · November 20, 2003 09:52 AM

Chris, I agree that the systems aren’t designed exactly to maximise life expectancy. But I think infant mortality is a pretty good test, since that really is a data point that is an explicit target. That the US has the worst numbers in the developed world (and by quite a distance) is rather striking, and a serious mark against its health care system.

I agree though that a health care system could go from efficient to inefficient (by these measures) by spending a lot of money on palliative care for those nearing death, and this might be an entirely appropriate use of health care resources. If there’s evidence that that’s exactly the difference between gov’t based and private based systems then the data isn’t as compelling as I try to make it look. I don’t think that it is what’s happening in the US though (it could be in Switzerland - they do have a small life expectancy gain for all their spending) but it’s certainly a possible explanation of the data.

Posted by Brian Weatherson · November 20, 2003 10:18 AM

This discussion reminds me of my favoured NHS money-saving solution.

If anyone has a less than 10% chance of surviving an expensive treatment regime, then the only care they should be offered on the NHS is a copious amount of morphine to make their final days more agreeable.

Instantly you have the cash to pay for the hip replacement type operations that will affect people’s quality of life for decades, and the doctors to carry them out.

Meanwhile, if anyone really wants to sell their house to pay for a treatment regime with a marginal chance of success, then they still can.

Posted by john b · November 20, 2003 11:58 AM

Here’s wild thought that just came to me…

There’s some evidence that more inequality is correlated with worse outcomes independently of the absolute level of income of the worst off in the relevant society.

Very unequal societies are likely to have more private medicine.

In other words (at least part of) the explanation of the correlation is that the bad health care outcomes and the health system type share a common cause.

Posted by Chris Bertram · November 20, 2003 12:07 PM

“The usual conservative response to nasty data like this is to insist the US “free market health care system” is subsidizing the rest of the world’s socialist systems. No actual evidence is provided, but there you go.”

OK maybe this is tangential, but there is one clear case where the US does subsidize the rest of the world’s socialist health systems - drug costs. Socialized systems limit the prices of drugs, the U.S. does not. Ergo it’s the U.S. consumer which is paying for all that research, while the Socialist systems (by pricing the drug just above its manufacturing cost) are freeloading.

The U.S. consumer does get one benefit - drug companies are more likely to do research on diseases which impact Americans strongly, such as (if this really is a disease) obesity. But other countries still do derive the benefit without paying their fair share of the development cost.

Posted by Andrew Boucher · November 20, 2003 01:04 PM

By the way the conclusion that I draw from this situation, is that the U.S. should limit drug costs too. Remark (for all of those nodding your approval) that this is the unilateralist solution. The multilateralist solution would be for the U.S. to get together with all the other heavily drug-consuming countries and negotiate before it does anything.

Posted by Andrew Boucher · November 20, 2003 01:14 PM

Could one problem be that the US government is almost uniquely inefficient among OECD governments in providing and spending anything, because the political system is designed to encourage log-rolling and rent-seeking directly through the process of legislation? If this were true there might be a case for keeping things as they are.

Infant mortality rates may be quite misleading. They are staggeringly high for the poorest 20% of Americans; the rest get infant mortality rates pretty close to the OECD average. The poorest 20% get paltry primary care spending — they get urgent care treatment instead, which is just as expensive but far less efficient. There is very weak ante and post-natal care for the poor.

One thing that doesn’t show up in the estimates above are the costs absorbed by healthcare consumers — the opportunity costs faced by patients and their relatives trying to a) figure out which insurance company to choose; b) how to get the insurance company to pay for treatment and also the cost of feeling insecure about prospective healthcare provision.
Antoher thing that doesn’t shwo up is the amount of healthcare spending on children that is done in schools — this all comes under education budgets, and I’m sure it far exceeds the normal amount that comes under these budgets in other OECD countries. SO the figures above almost certainly UNDERSTATE the real costs of the US system relative to other systems.

Does the US subsisidise R&D? Whenever I see that argument made it is supported by strident stipulation, not economic analysis. Even if it does, though, drugs have very limited impact on health outcomes (as does health care generally, as Chris sort of points out). Over the course of their similarly long lives Americans work about 20% more hours than Europeans; they live further from their families; they spend much more time doing sedentary things like driving; they consume more fast food and they spend less time with their friends (because they live further away from them both because of the high rates of geographical mobility and because of the population sparsity even within urban areas once you get away from the Eastern seaboard). Living healthily reduces health care costs and increases life-expectancy.

Sorry to rant — you’ve touched a nerve.

Posted by harry · November 20, 2003 02:11 PM

I have several objections to this— (2) and (4) below expand on what Harry said:

1. infant mortality and life expectancy are, at best, very rough proxies for the quality of health care provision. They certainly don’t correspond very well to the things that most individual consumers want from health care at time of provision (friendly and prompt service, choice of providers, etc). Maybe they correspond well to what you think people should want out of a health care system, but that’s another argument entirely.

2. There are lots of non-health-care factors that influence infant mortality and/or life expectancy— some examples would include crime rates, rates of unwanted pregnancies, diet and exercise habits. Some of these may be strongly correlated to government health-care funding for cultural reasons having nothing to do with the effectiveness of said funding. I am reminded of Milton Friedman’s famous exchange with a Scandinavian economist, which went something like:

Other guy: “In Scandinavian countries we have no poverty.”

Friedman: “Well, among Scandinavians in America we have no poverty either.”

3. There are also cultural factors influencing the spend rate on health care that have nothing to do with the efficiency of provision. US health care consumers, as a previous commentator noted, demand lots of really expensive end-of-life care that Europeans do not; they also are notoriously quick to sue health care providers, driving malpractice insurance costs through the roof. These things are largely due to an American cultural tendency to believe that:

(a) death is essentially optional;

(b) any death that happens while under a doctor’s care should be assumed to be the fault of the doctor until proven otherwise.

Changing to a more socialist system of health care provision would do absolutely nothing to change these cultural attitudes, and it is therefore doubtful it would reduce expenditures.

4. Averages are misleading. The US may well be better than other nations at providing health care to most people, and still have low average outcome scores because a small percentage of people have really horrendous outcomes. Whether you think this vindicates the effectiveness of the US system or not depends on how egalitarian you are, I suppose; but it’s still worth noting that the average doesn’t tell the whole story.

5. Less direct government funding does not equal less government involvement. The data given do not provide any basis for a comparison between free-market and government-funded health care, because no developed country today has anything remotely resembling a free market in health care. Certainly the US does not, and has not for decades; indeed in many ways the US government probably micromanages the provision of healthcare more than some single-payer countries do.

Posted by Nicholas Weininger · November 20, 2003 03:50 PM

I think it will be news to the NHS that there is a “tax line item” representing part of its revenues, unless I’ve misunderstood what a “tax line item” is supposed to be.

Posted by Nasi Lemak · November 20, 2003 04:09 PM

Taking the percentages of public/private financing and turning them into dollar figures (absolute expenditures) yields some more interesting relationships.

Absolute private expenditures correlate significantly more with health care costs as % of GDP (.80 vs .64)

Public dollar expenditures correlate much more with increased life expectancy (.67 vs .31 female, .75 vs .36 male) and with decreased infant mortality (-.58 vs -.21).

Posted by David Yaseen · November 20, 2003 04:52 PM

Like most of the ‘stats’ produced by the OECD, the picture is distorted to show that the US is doing everything wrong. Instead of using life expectancy as a measure of ‘goodness’(mostly due to good food and effective sewers), it would be more relevent to use the cancer survival rate (mostly due to good health care). Check it out - and you’ll want to move to the US! Besides, the French system really did itself proud during this summer’s heatwave - no? It was safer to be a GI in Baghdad than a pensioner in Paris! Mind you, that’s probably true no matter what the weather in Paris…

Posted by Sean O'Callaghan · November 25, 2003 06:16 AM
Followups

→ Fun With Health Research.
Excerpt: Brian at Crooked Timber compares health spending and statistics of OECD countries. His conclusion: [...] health care systems that are government run or funded tend to be cheaper despite being just as effective in every respect, and more effective in...Read more at Appalachia Alumni Association
→ Is Nationalized Health Care More Efficient?.
Excerpt: It's a common thought on the left and among liberals that national health care is better not only because it is universal, but also because it's a less expensive way to achieve the same health outcomes. (See, for example, Brian...Read more at The Cardinal Collective
→ Public and Private Health Care.
Excerpt: Brayden King notes that the Wall Street Journal is concerned about ever-rising health care costs in the United States. I’ve...Read more at Kieran Healy's Weblog

This discussion has been closed. Thanks to everyone who contributed.