Random Critical Analysis finds a connection.
The combination of the slowing efficacy of incremental health expenditures (flat of the curve spending) and the rising burden of western illnesses implies a potentially unambiguous negative relationship between income and life expectancy may arise amongst upper-income countries.
Read the whole thing. Case and Deaton describe opioid deaths as American exceptionalism. RCA emphatically disagrees, saying that greater national affluence is associated with greater rates of drug use. That strikes me as not really intuitive, because I am guessing that within a country, the relationship between income and drug use is negative, not positive.
By “western illnesses” he seems to mean lifestyle illnesses, including obesity and substance abuse. One thought that occurs to me is that health among Americans may have been hurt more than helped by the drop in smoking. Other forms of comfort are more harmful, so curbing one form does not help if other forms are substituted.
tldr; opioids shouldn’t be separated from opiates in this analysis
The relevant counter-factual is opiate addiction rates in Iran, due to the availability of cheap Afghan opium. The correlation between opioids and affluence occurs because opioids are manufactured by pharmaceutical companies and the legitimate market for opioids is health care. Opioids and opiates should be treated as a single substance category when measuring the impact of affluence.
I think you’re undercutting your own argument here. A simpler model may be that drugs addiction is high in Iran for the same reason it’s high in the US: drugs are more affordable than elsewhere. They’re just affordable for different reasons; in the US, because we’re really rich; in Iran, because drugs are really cheap.
I also don’t see how pharmaceutical companies are supposed to explain rates of heroin use in developed countries. Heroin, of course, is also a very expensive drug. It was much the same with cocaine as well back in the 80s and 90s. It hit developed countries harder because they could afford it.
I think you are partly making my argument, Mark. Factors other than affluence (e.g. price, transaction costs, risk, availability, “despair”) account for addiction rate differences between countries.
The affluence argument being made by “Random Critical Analysis” is based on opioid availability between countries. That is tautological since the availability of pharmaceutical opioids increases the more affluent a nation/health-care-system is.
If you look at the List of countries by prevalence of opiates use you see a mixture of affluence and some drastic differences in similar nations like Hong Kong and Macau. As Kling points out, there is also an inverse relationship between affluence and addiction within countries.
Focusing only on opioids distorts the bigger picture rather than clarifying it. Supply chains and distribution channels and government efforts to disrupt them seem to play a more important role.
You say this is tautological, but (1) many people would find this surprising; (2) I’m not arguing healthcare systems have nothing to do with it; (3) healthcare spending is very much tied to income levels; (4) the factors you mention are hardly disconnected from income levels either. Even if you wish to argue the effects of income are channeled entirely through healthcare provision (unlikely) and that there are other factors (not arguing it explains 100% of the variance!), you’re tacitly acknowledging this form of consumption is an (indirect) income effect. Significant positive relationships are also found for illicit drugs and overdose deaths, so it seems you’re reaching.
Incidentally, though I did find a significant income effect for opiates, the data are clearly spotty and probably aren’t terribly useful (see the notes: different methods, data sources, years, groups, etc). The figures I cited are likely to be more reliable and more useful for reasons relating to consistency, national statistical capacity, underlying data availability (illicit vs prescribed), intensity, and so on. Nonetheless, this is rather problematic for your position if you’re imagining a strong substitution effect and that these data are reliable….
The r-squared doesn’t need to be equal to one in order for the effect to be material.
Oh, I forgot to mention heroin. Mexican heroin is an opiate. Chinese fentanyl is an opioid. They are substitutes for one another. Chinese fentanyl base pills are also a substitute for oxycodone. The opioid/opiate distinction is based on data sources, not consumer behavior.
I’m not knowledgeable about health issues in general. I also don’t doubt the link between smoking and cancer. But, I’m confused how a country like Spain, where 1/3 of the adult population smokes (compared to the approx 1/7 in the US), regularly tops healthiest country index’s like the OECD or Bloomberg’s recent health index. Is smoking not the killer I’ve been taught it is? Is it more genetics than anything else? Was I stupid to quit smoking 8 years ago and trade it for a few extra pounds?
As I mentioned on my blog, the evidence suggests the peak mortality effects of smoking lag by several decades. That is, other things equal, the countries that suffer most from smoking today are those that smoked the heaviest 2-3 decades prior. Countries like Spain will presumably be paying a heavier price a decade or two down the road.
https://tobaccocontrol.bmj.com/content/21/2/96
https://ourworldindata.org/grapher/sales-of-cigarettes-per-adult-per-day?time=1955..2014&country=ESP+CHE+USA+ITA+FRA
“Other forms of comfort are more harmful”
It’s good to have new ideas and explore them, but I think you will have a hard time proving that many things (anything?) is more harmful than smoking. Did you have thoughts or evidence to back up this claim?
Sorry this was a response to Arnold but was accidentally placed in the wrong sub-thread.
I think you will have a hard time proving that many things [are] more harmful than smoking.
Robin Hanson argues that the net danger of smoking has been exaggerated.
– http://www.overcomingbias.com/2009/12/what-anti-smoking-evidence.html
– https://www.overcomingbias.com/2009/12/smoking-followup.html
Random Critical Analysis says below that you see peak mortality effects 2-3 decades after period of heaviest smoking. I think the studies Hanson cites look at shorter time frames than that, so maybe he would revise his opinion after looking at stats over a longer term.
*** reposting without direct links – auto-moderation issues ***
As I mentioned on my blog, the evidence suggests the peak mortality effects of smoking lag by several decades. That is, other things equal, the countries that suffer most from smoking today are those that smoked the heaviest 2-3 decades prior. Countries like Spain will presumably be paying a heavier price a decade or two down the road.
(copy and paste into browser)
tobaccocontrol.bmj.com/content/21/2/96
ourworldindata.org/grapher/sales-of-cigarettes-per-adult-per-day?time=1955..2014&country=ESP+CHE+USA+ITA+FRA
Dude, fantastic blog and excellent work. I for one think you’ve raised important questions that need to be considered more widely. Looking forward to spending many more hours reading your blog.
Thank you.
Well you can ask any doctor and I sure 80%+ would trade smoking for ten pounds. Or in my business, you can pay $40/month more for health insurance.
Also Japan smoking rates (19% to US 17%) has a long life expectancy than Spain. It is just Spain other health rates are higher than the US to cover higher smoking rates. (Accidents are also a big piece of US deaths.)
Think about it this way. If I am a drug kingpin in a third world country, where should I smuggle my drugs to maximize my profit. I suspect the first world countries, even with higher interdiction rates, would be more profitable due to their higher incomes.
This analysis is too ‘Hot Take’ to make any big conclusions here.
1) Of course Life Expectancy is going to diminishing returns as the human body is not meant much longer than 70 – 75. You can only cure cancer and heart disease so long.
2) I still see the big picture here that modern economies have longer life expediencies than non-modern economies.
3) There is still a wide disparities in states. The difference of Hawaii and California to Mississippi and West Virginia is over five years. (Asian populations?) Why not look at this as well? I still think the simplest reason for this disparity is urban states have better access to Emergency medicine and rural states are losing a lot of clinics.
4) I think the real economic issue is dropping in class. The lifestyle in the Rust Belt is better than say 75 years ago, but a lot of working class are dropping in class or two classes which is very hard to do.
Anyway, we still have Japan who is affluent modern nation who does have the highest life expectancy.
Note, in viewing the states, I don’t know how Puerto Rico does well here.
You’re not going to get any argument from me there. Of course, the United States isn’t usually directly compared much less developed countries either.
I’ve explored such issues before, but my blog post was already longer than many peoples’ attention spans and some of these issues are politically touchy.
I think regional and ethnic group patterns are informative, but intra-national differences in healthcare provision are likely to play a very small role at best. If you look at county-level data for indicators of healthcare provision like physician density, % uninsured, preventative care indexes, and so on, the coefficients are usually very close to zero (even from a bivariate perspective) whereas obesity, smoking, homicides, drug abuse, and the like are strongly associated with high mortality rates. Most of the areas with high mortality are not *that* rural and have healthcare that looks very much like the rest of the country (at least they do in observable dimensions).
Much of the regional clustering in these behavioral dimension (obesity, smoking, drug abuse, etc) is consistent with shared deep roots and is generally not well explained by socioeconomic indicators like income levels, education, and so on. Whites in Appalachia, the lower midwest, and parts of the deep south (historically in the upland regions) share common settler origins and indicators of ancestry seem to explain much even net of controls for things like income, education, state fixed effects, etc. Their forebears largely came from the periphery on the United Kingdom and these source regions seem to do much worse in the UK as well…..
I find changes in relative status, employment, and the like to be plausible contributors, but mostly I suspect these are secondary to long standing differences. That is, my mental model looks more like income + deep roots (-> human capital, social capital, etc), possibly with something that looks like an modest interaction effects when viewed on balance (as a function of income). That is, given an environment that increasingly allows people to make poor lifestyle decisions (overeating, drugs, risky sex, etc), some groups of people are less able to resist whereas others are relatively more able. Those that do a relatively good job avoiding such temptations are also more likely to benefit from some of the cutting edge medicine that might incrementally extend life in old age.
“Asians” (read: mostly East Asians) in most states in the United States live quite a bit longer than white Americans and Europeans in most high-income countries in Europe. They also typically have relatively low BMI by western standards and avoid most of the lifestyle issues that seem to beset many Americans. I don’t think that’s a coincidence (though genetics may play role here even conditional in such factors).
In terms of states, it is no accident that two highest life expectancy states, HA & CA, are also two highest portion of Asian-American populations.
Probably the real ironies in the United States the last generation are Hispanic-Americans are living longer than Euro-Americans. The highest growth in life expectancy is Puerto Rico of all places. So simple take is many in the Hi-American community still feel like they are moving up in economic class as compared to more Euro-Americans in economic depressed areas.
I’m not sure why this is unintuitive; it actually makes sense to me that poorer people in wealthier countries would tend to do more drugs. A few possible reasons: 1) To some extent, drugs may be a palliative for low social status; so poor people anywhere may be more inclined to do drugs, but poor people in rich countries can better afford them. 2) drug use is largely a matter of poor impulse control (and/or other psychological traits) that tend to correlate with relative poverty within a country, but not between countries. In fact, poor people in poor countries are more likely to be poor for other reasons (and less able to afford drugs) than poor people in rich countries. Finally, 3) drug use is both expensive and poverty inducing, so rich countries have more users, and users tend to be poorer.
Overall, the correlation with wealth between countries and negative correlation within countries actually seems to have a good number of intuitive explanations imo.
On Bloomberg radio today, someone from Oramed was talking up progress they are making in getting insulin pills to market. Apparently substituting pills for injections may have considerable cost savings.
Diabetes does seem linked to affluence and the growth in diabetes spending is staggering. Especially with “pre-diabetes” becoming a big thing. From elsewhere:
“Spending on diabetes and pre-diabetes is currently $322 billion per year, up from $245 billion in 2012 and now accounts for one-fifth of overall healthcare spending in the U.S.1 From 2015 to 2016, U.S. spending on diabetes related medications increased from $43.9 billion to $51.5 billion.”