Sentences to ponder

Overall, our findings suggest that correlations observed in affluent, developed countries between (i) wealth and health or (ii) parental income and children’s outcomes do not reflect a causal effect of wealth.

The paper is here, commentary from Arnold Kling is here.


It seems that there is a problem with the link to the paper :)

Here's a fixed link:

Swedish lottery players?

Thanks Arjun :)


...and never the twain shall meet? Lawrence Summer and Lant Pritchett authored the most cited paper on health economics ever issued by World Bank staffers (see the Adam Wagstaff overview that suggests that neither of these authors can, or would, claim to be a health economist). Their title, "Wealthier is Healthier," tells their story. Good health economists like David Cutler, Angus Deaton, Adriana Lleras-Muney hold the opposite view (as do I and many others).

Thanks for bringing to our attention one more empirical piece on this important topic. If John Snow had not gotten the pumphandle removed, if scientists had not identified the causes of infectious and communicable diseases in closely-spaced cities in the late XIX century, if propinquity had not permitted exchange of information and productivity growth, would we be so rich as we are today?

I think not, but what do other MR readers think?

Unless you're an ardent blank-slatist, it's pretty obvious that the link between health and wealth is largely genetic. Waist-to-hip ratio is one of the primary determinants of longevity. Men strongly prefer women with low WHRs. Rich men have a natural advantage in the mating market. Upper class children receive their father's genes for numeracy and conscientious, and their mother's pear-shaped genes. Hence they go on to live long, rich lives.

What about the bumps in people's skulls? How do they compare in importance to WHR?

Umm, actually pretty gosh darn important actually...

"...based on 37 samples across 1530 people, the population correlation was estimated at 0.33...For external head measures, Vernon et al. (2000) reported the population correlation between head size and intelligence to be 0.19. Nguyen and McDaniel (2000) reported population correlations from 0.17 to 0.26 for three different sub-categories of external head size measures. Studies assessing the correlation between in vivo brain volume and intelligence are more rare. Vernon et al. (2000) reported data on 15 such correlations and obtained a population correlation of 0.33. Nguyen and McDaniel (2000) reported the same population correlation based on 14 correlations. Gignac, Vernon, and Wickett (2003) reported data published in 2000 or earlier with a mean correlation of 0.37."

Science is scary, albert. You're right to flee from it. Magical thinking for all!

Totally not trying to be a dick; just curious:

Do you believe in climate change?

What was your dissertation topic in?

Believe in science when it coincides with your priors, appeal to authority when called out on bullshit. Albert, do you vote Democrat?

I am not a blank-slateist but is not obvious to me that most health is genetic. Nor do I think that there are genes for numeracy and conscientious.

These things are likely to be partly genetic but mainly taught. Becoming a Mormon has an impact on your health. So does being divorced. Or the child of divorce.

It is more likely that good habits and a strict up-bringing leads to wealth.

Clean living helps, but genetic effects have by far the largest effect on longevity.

"Research also suggests that there is a clear link between living to 100 and inheriting a hyperactive version of telomerase, an enzyme that prevents cells from ageing. Scientists from the Albert Einstein College of Medicine in the US say centenarian Ashkenazi Jews have this mutant gene.[14]

Many centenarians manage to avoid chronic diseases even after indulging in a lifetime of serious health risks. For example, many people in the New England Centenarian Study experienced a century free of cancer or heart disease despite smoking as many as 60 cigarettes a day for 50 years. The same applies to people from Okinawa in Japan, where around half of supercentenarians had a history of smoking and one-third were regular alcohol drinkers. It is possible that these people may have had genes that protected them from the dangers of carcinogens or the random mutations that crop up naturally when cells divide.[15]

It is well known that the children of parents who have a long life are also likely to reach a healthy age, but it is not known why, although the inherited genes are probably important.[17] A variation in the gene FOXO3A is known to have a positive effect on the life expectancy of humans, and is found much more often in people living to 100 and beyond - moreover, this appears to be true worldwide.[18]"

The outliers in any field are likely going to be there because they hit it big in all categories.

Exactly, which is why so many extremely long-lived people having bad habits is demonstrative of lifestyle's de minimum role relative to genetics.

Analogously imagine if we were examining the characteristics of the fastest cars every made. We'd expect that rather than simply being good at one thing, they'd have nearly perfectly engineered components in almost every category. Their engines would have to be super-charged, their transmissions expertly engineered, their suspensions tight, their frames delicately lightweight yet capable of handling extreme power. A car that was well-crafted in in some major categories, but lacking in others, might be a fast car, but definitely not one of the fastest ever made.

But imagine many of the fastest cars in the world, had some sub-standard component, that common wisdom purports to be extremely important. For example what if everyone assumed having wide tires and strong axles were extremely important to automobile performance. Yet we see many McClarens, Lamborghini and Bugattis with narrow tires and cheap axles.

Well it'd be one thing if we saw it on fast, but non-record setting cars. That could be explained away by compensation along some other dimension. But if tires and axles were critically important, then being deficient in them, would assure that a model would never be able to push the boundaries of what's possible. The only conclusion we could reach is that tires and axles may play some role in auto performance, but most likely a small role role compared to the other major factors.

Regular alcohol consumption is health -promoting

Even under the most generous assumptions, alcohol has no where near enough of an effect to make up for smoking 2 packs a day.

i'll let you know how my experimenting goes in 30 years (or not)


By far age is the strongest correlate with health. Oddly reverse age, is even better ;-)

Anyway, I thought your brand of social Darwinism died with the Holocaust.

Godwin's Law. You lose

Well, time to repeat Prof. Cowen's word when it comes that subject - 'ultimately the Nazi connection will be seen as a bump in the road'

Why do you take this as a statement of approval on Tyler's part? Tyler made a neutral statement describing how he foresees the future unfolding.

Do you really have to ask why, Ricardo?

"Anyway, I thought your brand of social Darwinism died with the Holocaust."

Hitler killed the Jews, therefore health is not strongly affected by genetics. Thanks, your logic is impeccable.

Ahem, that is not even close to a valid methodological design, and very little can be extrapolated to the general population from those findings. Their sample, being constrained to lottery players, is highly biased and does not even come close to reflecting the impacts of long term multi-generational wealth. Conditioning on lottery players introduces a large source of self-selection bias, as established wealthy individuals have much less incentive to play lotteries to those suffering from long term poverty.

Somebody needs to go back to experimental design school...very naughty indeed!

Looks like you didn't bother to actually read the paper. The author's already analyze and control for the differences between lottery players and the general population. (Table 4). Furthermore the authors find no health impact for children born *after* their parent's reward. If children born into multi-million net worth households don't constitute "established wealthy individuals" then the term might as well be meaningless.

I don't think the differences they can measure and control for are necessarily fine-tuned enough to detect an underlying behavioral difference that might make someone both (1) a lottery player and (2) the type of person whose new wealth fails to improve their and their children's health outcomes.

For instance, I am married and college-educated, but have also played the lottery in the U.S. despite knowing the terrible expected returns of doing so. I am also fat despite fully understanding the terrible expected health outcomes from being so. There may be something more fundamental about me that can't be teased out by knowing that I am married and well-educated, that makes me both more likely to play the lottery (and hence to win it) and more likely to not have improved health outcomes as a result of increased wealth. I may also pass this trait on to my children.

I looked at the paper, especially the discussion of the representativeness of the lottery players and at Table 4 which lists a lot of the 'usual suspects' demographic features and how they compare between lottery players and Swedes as a whole. They could control for age (lottery players are older on average).

There's no attempt to control on 'willingness to play a lottery'. They do address the issue a bit, pointing out that the lotteries were very popular and one kind of lottery (the savings account that pays interest in lottery chances instead of cash) was so popular that it constituted about half of all savings accounts, so lottery players can't be that different from all Swedes. I'm not persuaded that's good enough.

I agree with Aaron Sheldon. This is a problem with the paper, as there is likely a connection between 'willingness to play a lottery' and 'risk-taking' that might impact health, etc.

I am one of the authors of the paper (Lindqvist).

It is important to distinguish between i) the representativeness of the sample under study (external validity) and ii) the validity of the estimate for this particular sample (internal validity).

Within the group of people who participate in the lotteries in our sample, wealth is randomly assigned (conditional on specific "cell fixed effects"). This implies that we obtain an unbiased estimate for the effect of wealth on health for people who play the lottery. The fact that people self-select into playing the lottery does not cause of bias here. However, it is true that one cannot export our estimate for lottery players to the population as a whole without making further assumptions. Given that our lottery samples comprise such a large fraction of the Swedish population, and the small differences in terms of observable characteristics between the lottery sample and the population as a whole, we argue that the effect of wealth in the sample of lottery players is likely to be similar to the effect for the population as a whole. But it is true that though they do not have to be exactly the same.

Is there the possibility of substantial variation in how much individual players wager, and hence in how likely they are to win? It may be true that "lottery players" are fairly representative of the general population, while "heavy lottery players" are not. From a quick read of the section on each lottery it seemed like players could purchase additional tickets (though I'm not sure on the Prize-Linked-Savings accounts).

There is indeed some variation in the amount of lottery tickets held by each individual. This does not cause bias since we compare outcomes of lottery players with the same number of lottery tickets (but different prizes). However, it is true that the representativeness of the sample could be affected if people who buy many lottery tickets (and consequently are more likely to win large prizes) differ markedly from the population. Yet we find no evidence of this: weighting each observation by amount won does not change observable pre-win characteristics in a meaningful way.

Lottery players could always buy more lottery tickets at some later date. However, the comparisons we make are between people with the same number of lottery tickets, at the same lottery, at the same point in time. Lottery tickets bought at a later date does thus not bias our Estimates.

In fact the authors themselves discouraged such broad generalizations as you have conjectured here. The authors in the discussion specifically qualify that the findings are only applicable to understanding the impact of wealth shocks, not the long term impact of multi-generational wealth.

Tyler didn't conjecture anything. That's a direct quote from the paper.

This must be one of the "Straussian" posts.

I think Strauss was generally a bit more subtle than this...

I'm sympathetic to some of the conclusions and problems with trying to improve outcomes, but I'm not sure this paper really shows us anything useful...

I really don't see how you can extrapolate from Swedish lottery winners to all affluent, developed countries. Isn't Sweden already distributing a lot of money and services not through lotteries but to everyone?

We point out in the paper that an important difference between Sweden and the US is that Sweden provides universal health care (almost) free of charge. It is therefore unlikely that lottery wealth shocks could affect access to care in a meaningful way. However, many of the mechanisms for the health-wealth gradient that have been put forth in the literature are unrelated to direct access to care (e.g., stress, life-style). A reason for this is that the health-wealth gradient show up reliably in countries with very different health care systems. In the paper we show that the gradient between wealth and mortality are similar in Sweden and the US. Consequently, it is not enough to refer to health care access to explain the gradient, and a Swedish setting is relevant for testing the alternative mechanisms.

My parents were all, "Hey, our high incomes provide you with food, shelter and medicine." and I was like "Yeah? What's your confidence interval?"

Health correlates with wealth. A cautious, hard working, healthy, self=controlled person has a better chance of making money and those same attributes will help a person live longer.

For a control group how about a few groups who got money, tutors, education, opportunity, name recognition etc. but were big risk takers:

According to a study cited by Sports Illustrated in 2009, 78 percent of former NFL players are bankrupt or undergoing severe financial stress within two years of retirement from football, and 60 percent of former NBA players are bankrupt within five years of retirement.Jan 22, 2014

An estimated 60 percent of N.B.A. players are broke within five years of retiring, and 78 percent of N.F.L. players are bankrupt or under financial stress because of joblessness or divorce within two years, according to a report by Sports Illustrated in March.Oct 5, 2009
The Wall Street Journal reports that a team of UK scientists observed nearly 1,500 artists from across a wide variety of musical genres over a 50-year span, and were startled to discover that nearly 10 percent of them passed away during the study.
But before you put down your guitar out of fear for your health, don’t worry — it wasn’t the music. In fact, as the WSJ puts it, “Nearly 39% of those deaths were caused by factors related to violence or alcohol and drug intake.” This conclusion wasn’t lost on the team conducting the study, who named the paper summarizing their findings ‘Dying to be Famous.’

+1 to this post.

I love the progressives coming out of the woodwork to:

a) attack Tyler, even though he said nothing

b) attack the authors for poor design without actually reading the paper

c) say that no matter what the paper says they know the truth

I'm actually shocked that they haven't threatened the authors yet. Interesting. My guess is that it is about to happen. Never change proggers.

why are you assuming all critics are progressives? loool .... Also this a blog, if everybody agreed with the paper and with Tyler, then the title should be changed to pravda revolution ;) ... Also it seems you have poor knowledge of how science works (no criticism = no science), which might be the reason why your posts are so one-sided (just another right wing pro-market troll...nothing new here)

"no criticism = no science"

The commentators were not criticizing the paper, as they had not even read it. Just the conclusion they did not care for.

The commentators did criticize Tyler, when he made no argument at all, so they did not read the post.

Should I conclude they are just another left wing anti-market troll, because they chose not to understand what they criticize? And what does this post have to do with anti or pro market anyways? Did you not bother to read the post?

Since he posted the article without any comment, then I would say he implicitly endorses the claim... so he is making an argument. "
"Should I conclude they are just another left wing anti-market troll, because they chose not to understand what they criticize? And what does this post have to do with anti or pro market anyways? Did you not bother to read the post?" That's why i am asking "why are you assuming all critics are progressives? loool ", Apparently you didnt read my comment either ;)

I'm a rabid redistributionist yet love this post. Basically: national healthcare system baby! Reduced cost but no significant health tradeoff.

Does anybody ever study outcomes for only children versus kids raised in larger families? All else being equal, only children have a lot more financial resources behind them. I was an only child, and even though my parents would be only lower middle class by Robert D. Putnam's measure, life was comfortable relative to kids at school with, say, 3 siblings. But what the long term effects of having that kind of cushion is, I can't say.

On the flip side having siblings gives one a nearly inalienable, highly-trusted built-in peer network. How many people start companies with their siblings, get employed at their sibling's firm or enjoy legacy status when applying to their sibling's college? And as parents age and need to be cared for, sharing the burden between four or five children is much easier. Having lots of aunts and uncles around makes childcare much easier through pooled effort. A lot of research shows that mid-to-late life wellbeing is highly influenced by the number of close friends a person has. A brother or sister is pretty much guaranteed to be a close friend for life.

There is a lot of literature on this subject which tends to find a negative correlation between family size and educational outcomes however with proper corrections (parental characteristics including educational attainment of parents) the effect becomes insignificant indicating family size does not seem to have any effect. Although you can find papers arguing the opposite i suspect that the better the study the smaller the effect. This would be in line with other research in behavioural genetics.
"Like many studies, this paper finds a negative correlation between the number of children and educational attainment, and this is especially true for the United States. This negative correlation declines however when control variables like parental education and birth order dummies are added. This signals that the observed negative correlation might not be causal. The IV results indeed are no longer signiÖcantly negative, but positive and insignificant. Although the standard errors are not small, these results indicate that exogenous variation in the number of children does not have a significant negative effect on the educational attainment of a child.
Monique de Haan, Birth Order, Family Size and Educational Attainment

A study found that older siblings had higher IQs if their younger sibling was born more than two years after they were. That would seem to apply to an only child as well.

Good points.

Re: the Wealth and Health question:

Race or class versus race and class: mortality differentials in the United States

and pretty much all the work by Vincente Navarro of Johns Hopkins Bloomberberg School of Public Health

See, for instance,
"Inequalities are Unhealthy"

Modern health care=survival and reproduction of inferior genes. Absolutely no argument about that. The consequence of this is the evolution of the human is now in a retrograde phase, lots of genetically inferior individuals are surviving to produce more like themselves, people with a predisposition to diabetes, heart disease, maybe even cancer and who knows what other afflictions. In a couple of hundred years everyone will have a handicapped parking plate and institutionalized health care will be the most important thing in the country. Of course, it may well have been taken over by the healthy, vigorous members of the supposed culturally backward third world by then.

Francis Galton is that you?

He's not even wrong.

Not even wrong? Humans have literally put the kibosh on natural selection. If you believe in evolution by natural selection, the consequences are obvious. If you believe in evolution by natural selection, but dismiss any discussion of genetic superiority/inferiority, you are full of cognitive dissonance.

Natural selection does not differentiate between "human selection" and "really truly natural" selection.

In Sweden, with a majority of government funded healthcare, citizens already have access to quality health coverage financed by taxes. This system ensures the well-being of the average patient thereby promoting a lifelong regimen of regular preventative medicine. As this socialized system is already the norm and every citizen is privy to this, then adding wealth on top of this system would not suggest that a lottery winner would all of a sudden be able to afford healthcare they previously could not. Is this completely reasonable?

This is correct: Winning the lottery in Sweden does not imply that you significantly increase your access to health care. This is an important difference compared to, say, the US and something we discuss in the paper. Yet the health-wealth gradient shows up across a wide range of countries with different health care systems, and definitively in Sweden as well. We show in the paper that the gradient between wealth and mortality is similar in Sweden and the US. Many theories of the gradient therefore invoke mechanisms that are unrelated to health care (e.g., that wealth may reduce stress which in turn improves health). The case of Swedish lottery winners is allows us to test the relevance of these alternative mechanisms.

Does the United States not approach the level of universal health care as Sweden? The indigent have medicaid, the elderly have medicare, the working largely have employer group policies. The vast majority of US citizens have access to some baseline health care - how does this baseline compare to Sweden's?

There is a difference between Sweden and the US in the sense that there are no people in Sweden without coverage. But I agree that the difference should not be overstated. And, as I wrote above, there is a strong gradient also in Sweden despite universal health care coverage.

Comments for this post are closed