The median cost of healthcare as a percentage of GDP for Australia, Canada, France, Germany, Japan, Norway, Sweden and the UK is 9.1%.
In the United States it’s 75% higher, at 16%. No wonder so many are concerned about the cost of healthcare in the United States.
Among these countries, with the exception of Norway, the United States has the highest GDP per capita yet a significantly larger population. The economies of scale explanation isn’t appropriate in this situation, but some might expect it to apply. Clearly it doesn’t. The wealth of a nation (if you want to judge wealth by GDP, or GDP per capita) doesn’t drive down costs of healthcare. There’s actually no statistically relevant relationship between the two.
With the exception of Norway and the United States, these countries spend less than $4,000 per person, per year on healthcare, with a median of around $3,300. Norway, on the other hand spends just over $5,900 per person. The United States spends more than double the median, at nearly $7,300 per person!
Why is the United States spending so much on healthcare compared with other industrialized democracies? Are people living longer? Nope. In fact, the U.S. has the lowest life expectancy among these countries. Also, the infant mortality rate is 75% higher in the United States than the median of the other nations. Something else must explain it, then.
There’s actually a really simple explanation – at least for a significant portion of why the total cost of healthcare in the United States is so far ahead of the rest.
Those who understand some basic statistics will likely be floored when they see the numbers. I’ll start with those, and then give a brief explanation of what they mean.
If you take the healthcare costs as a percentage of GDP and the percentage of healthcare costs paid by government, the coefficient of determination between these variables, or R2, is 0.73.
Now, for those of you who didn’t crap when you read that, let me explain what it means. That means that 73% of the variation in the percentage of GDP that goes to healthcare is explained by the percentage paid for by the government. That’s a whole lot of an explanation!
Additionally, we need to see the direction of influence. The Pearson’s correlation is -0.855. If you are in the social sciences, you probably aren’t used to seeing such high correlations, especially with something as seemingly complicated. It shouldn’t be this simple, but it appears to be. What this means is that the higher percentage the government pays toward total healthcare costs, the less people (and the government) spend on healthcare – and the impact is very powerful. If the government covers more of the costs, the costs for everyone – for citizens and the government itself – go down.
It makes some sense. As an individual consumer, any voting you do with your dollars is a drop in the bucket. (Actually, more like a molecule in a drop in a bucket.) When a government is spending a supermajority of the dollars toward healthcare costs, the economic influence alone is massive. But governments have much more influence beyond that, both in terms of incentives and regulations. It does appear that privatization in healthcare leads to significantly higher costs.
Things get a little more creepy when you look at some other measures. Let’s look at the relationship between the percentage of healthcare paid by government with infant mortality. In this case, the R2 value is 0.696. That means that 70% of the effect on infant mortality is explained by the percentage of healthcare paid by government. That means the U.S. government, which pays 45.4% of healthcare costs in the U.S. (the other countries cover a median of 80%), is significantly contributing to a lot of infant deaths by not covering a larger percentage of costs.
The bottom line: if we want healthcare costs to go down in the United States, the simplest solution seems to be for the government to begin paying for a significantly larger share of the costs. I also appears that this question is not merely a matter of monetary costs, but one of life and death.
I didn’t need to spend much time to figure this out. All I needed was some data from Wikipedia:
Health Care System, Cross-Country Comparisons: – The data source here is http://www.oecd.org
List of Countries by GDP: for this I used the IMF data from 2010
List of Countries by Population
I did the statistical work in Excel, using simple statistics functions. This is me playing DIY economics. You should try, and see what you come up with.
|
Country |
Life Expect. |
Inf. Mort. |
Phys. per 1k people |
Nurses per 1k people |
Per capita expend. on health |
Health Costs as % of GDP |
% of gov’t rev. spent on health |
% of health costs paid by gov’t |
GDP PPP |
Population |
| Australia |
81.4 |
4.2 |
2.8 |
9.7 |
3137 |
8.7 |
17.7 |
67.7 |
39,764 |
22,724,616 |
| Canada |
81.4 |
3.9 |
2.3 |
9 |
3895 |
10.1 |
16.7 |
69.8 |
39,171 |
34,605,000 |
| France |
81 |
4 |
3.4 |
7.7 |
3601 |
11 |
14.2 |
79 |
33,910 |
65,821,885 |
| Germany |
79.8 |
3.8 |
3.5 |
9.9 |
3588 |
10.4 |
17.6 |
76.9 |
36,081 |
81,751,602 |
| Japan |
82.6 |
2.6 |
2.1 |
9.4 |
2581 |
8.1 |
16.8 |
81.3 |
33,885 |
127,950,000 |
| Norway |
80 |
3 |
3.8 |
16.2 |
5910 |
9 |
17.9 |
83.6 |
51,959 |
4,972,400 |
| Sweden |
81 |
2.5 |
3.6 |
10.8 |
3323 |
9.2 |
13.6 |
81.7 |
38,204 |
9,440,588 |
| UK |
79.1 |
4.8 |
2.5 |
10 |
2992 |
8.4 |
15.8 |
81.7 |
35,059 |
62,435,709 |
| USA |
78.1 |
6.7 |
2.4 |
10.6 |
7290 |
16 |
18.5 |
45.4 |
46,860 |
312,362,000 |
While I’m sympathetic to what you did there and where you’re going with it, this needs to be salted with the old “correlation does not equal causation” reminder.
In effect, what makes you so sure it’s who’s doing the paying that’s the reason for the strong correlation? There are many other third variables that also differ between countries with more government sponsored healthcare and those that lack it.
Let’s take a good wide contrast to illustrate the argument, the USA vs the Netherlands. The Dutch have a super strong social safety net including healthcare and all kinds of other good stuff that people need to survive and thrive. So, for one thing, the gov’t providing healthcare is correlated with the gov’t providing a decent level of all sorts of other things like, housing, education, and nutrition. As a result, people are healthier, also less stressed, competitive, and pissed off, also resulting in better health. They also don’t drive nearly as much, preferring to bike and walk and making it feasible to do so, and they don’t eat the same ridiculous portions or types of food as Americans.
Taxes are a lot higher in the Netherlands and it’s a lot harder to create a big wealth disparity between yourself and others. When people manage to, it’s not as big a difference as it would be over here, because the lows arent are low and the highs aren’t as high. And when people do manage to leave the pack, there is not so much of a osical reward. The Dutch don’t look up to people who seek to show off with luxury goods – it’s more about a better standard of living for all.
In that kind of context, there is less incentive for doctors or other healthcare providers to gouge people for their services, because it’s just going to get them taxed at a higher rate, and no one really has to worry about going hungry and homeless and starving in the streets unless they essentially choose to have that kind of life.
So while I would concede in this comparison that there is a correlation between how much the gov’t pays for healthcare, costs, and health indicators like infant mortality, on a causal level I would say it has much more to do with an attitude or culture that is expressed on a national level – of looking out for people, providing a community, not leaving people on their own to slip through the cracks, not fetishizing luxuries, and so on.
thus I don’t think the US overnight agreeing to pay for most of citizens healthcare would itself lead to those positive changes. Our culture needs to find its heart and soul first. Providing a better safety net would just be one reflection of an attitude of not idealizing greed and gluttony, but of preferring community, which is the true provider and indicator of people’s and societies’ health.
Unfortunately somehow in the US, this attitude of greed has taken deep roots among many people, even though data and experience shows it doesn’t serve them, satiate them, or make anyone particularly happy or healthy. This is all founded on an underlying fear – an insecurity – ultimately, a lack of faith in anything – leading to a lack of compassion and attitude of ‘me first’.
I don’t have an R squared to prove all of this, but this is my theory based on years of observation and reflection…
I could construct a model with more data to control for the things you’re suggesting. Perhaps at some point I will. A multivariate regression and chi-squares with controls would be able to test this. I’d be willing to bet that those items you mentioned are not intervening variables. I haven’t even consulted the literature on this, directly, though.
I do know that in the United States, the primary cause for increases in healthcare costs in the last decade are new technologies, followed by an increase in elderly patients receiving more expensive care. A congressional budget committee published a pretty exhaustive report on this recently.
Perhaps the stronger argument is:
Being in the United States makes healthcare cost more. Not being in the United States makes healthcare cost less.
Thanks for reading and the input.
The biggest driver of rising health care costs in the US is the lack of regulation of cost setting.
In Japan, for example, prices for all aspects of health care are set by a central board that meets every four years. They set and reset prices for everything from a single aspirin to the paycheck that you cardio-surgeon gets. The effect of this is that not only are costs contained, but prices are fixed no matter what hospital or clinic you go to, eliminated price gouging and price competition. Germany and other European countries also use a similar model.
I really don’t think that the “percentage of health care paid by the government” explains anything anything about costs. That relationship may hold statistical weight, but doesn’t imply any notion of causality. In fact, it oversimplifies the entire conversation.
The problem here is the wild-west, market model of health care that we use. It may work well to improve quality of care for those that can afford it, but does little good at containing costs as health care doesn’t operate like regular markets. One can’t just go somewhere else for a cancer treatment, and cost shopping is impossible. In the end, health providers can just charge whatever the hell they want giving them an effective monopoly on prices.
A heavily regulated system could control costs and would not result in declines in quality of care, as the Japan model shows.
Do you think that a government that pays over 81% of healthcare costs (Japan) or one that pays 45% (USA) is more inclined to set maximum prices? These data hold statistical weight not because this alone explains causation, but that the relationship between the percentage paid by government to a host of other factors. It may be the best single measure to capture things such as the broader contribution of government to the social safety net (as Chris mentioned) or the fixing of costs (as you mentioned), or a plethora of other issues. I think this figure alone is indicative of a broad range of other governmental behaviors and styles. I challenge anyone to find any other single measure with this high a correlation to the cost of healthcare.
I don’t know man, I just downloaded the data and found that the Pearson corr. coeff. between percent of GDP on health care and percent of health care expenditures by the public for 2008 among 34 countries was .0008. A plot of the two revealed nothing as well. Perhaps I’m doing something wrong, but that’s what I get.
Regardless, the United States is an outlier in percent GDP spent on health (16%), meaning that it’s not even comparable to the others, which only range from 5.8% to approximately 12%.
Removing the US doesn’t help much, either. I still only get a corr. coeff. of .08.
From what I have here, there’s no relationship between the percent of GDP spent on health care and the percent of health care costs borne by the public.
I don’t mean this as an attack and I’m open to correction.
Pete
I have included the spreadsheet in the post now so you can make sure we’re looking at the same numbers.
I was using the CORREL function, rather than the PEARSON function. Still, the correlation is comparable. If I had access to SPSS, I’d be able to do some more robust statistical work on it without spending much time. We must be looking at different data.
We’re looking at the same data, but I have 23 more countries than you do. I respect what you’re trying to do, but leaving out so many countries drastically reduces the variability of your data. How can you include so many European countries and leave out, for example, Ireland, Austria, Italy, Denmark, and Spain? Or include Japan and not South Korea? Australia and not New Zealand?
Also, Looking at some of your variables here, some of your correlation coefficient appear to be driven entirely by outliers. Removing the US from your dataset brings it somewhat back to reality, but again, you are dealing with a sample of 8. Not much to go on.
Including all the countries from the OECD data drastically changes the story.
I don’t know, perhaps you are correct, but the data doesn’t support it.
Again, my intent is not to attack. I am genuinely interested in what you are trying to do here.
Pete, thanks for the feedback. I didn’t attempt to find a more complete data set. I just grabbed was was accessible to me with limited time. If the model I used is “driven by outliers,” then a linear regression is probably not appropriate. I’d need to apply a curvilinear model to see if that more appropriately fits. That’s a lot of calculation, and something I’ll want to revisit at some future point since I don’t have access to a statistical package to do it for me.I appreciate your thoughts on this.