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 per capita in US$ (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.