Monday, June 13, 2016

Lottery Winners Don't Get Healthier

Alex Tabarrok at Marginal Revolution had a great post last week, Lottery Winners Don't get Healthier (also enjoy the url.)
Wealthier people are healthier and live longer. Why? One popular explanation is summarized in the documentary Unnatural Causes: Is Inequality Making us Sick?
The lives of a CEO, a lab supervisor, a janitor, and an unemployed mother illustrate how class shapes opportunities for good health. Those on the top have the most access to power, resources and opportunity – and thus the best health. Those on the bottom are faced with more stressors – unpaid bills, jobs that don’t pay enough, unsafe living conditions, exposure to environmental hazards, lack of control over work and schedule, worries over children – and the fewest resources available to help them cope. 
The net effect is a health-wealth gradient, in which every descending rung of the socioeconomic ladder corresponds to worse health.
If this were true, then increasing the wealth of a poor person would increase their health. That does not appear to be the case. In important new research David Cesarini, Erik Lindqvist, Robert Ostling and Bjorn Wallace look at the health of lottery winners in Sweden (75% of winnings within the range of approximately $20,000 to $800,000) and, importantly, on their children. Most effects on adults are reliably close to zero and in no case can wealth explain a large share of the wealth-health gradient:
In adults, we find no evidence that wealth impacts mortality or health care utilization.... Our estimates allow us to rule out effects on 10-year mortality one sixth as large as the crosssectional wealth-mortality gradient.
The authors also look at the health effects on the children of lottery winners. There is more uncertainty in the health estimates on children but most estimates cluster around zero and developmental effects on things like IQ can be rejected (“In all eight subsamples, we can rule out wealth effects on GPA smaller than 0.01 standard deviations”).
(My emphasis above)

Alex does not emphasize the most important point, I think, of this study.  The natural inference is, The same things that make you wealthy make you healthy. The correlation between health and wealth across the population reflect two outcomes of the same underlying causes.

We can speculate what those causes are.  (I haven't read the paper, maybe the authors do.) A natural hypothesis is a whole set of circumstances and lifestyle choices have both health and wealth effects. These causes can be either "right" or "left" as far as the evidence before us: "Right:" Thrift, hard work, self discipline and clean living lead to health and wealth. "Left:" good parents, good neighborhood, the right social connections lead to health and wealth.

Either way, simply transferring money will not transfer the things that produce money, and produce health.

Perhaps the documentary was right after all: "class shapes opportunities for good health."  But "class" is about more than a bank account.

Also, Alex can be misread as a bit too critical: "If this were true." It is true that health and wealth are correlated. It is not true that more wealth causes better health.  The problem is  not just "resources available to help them cope."

Why a blog post? This story is a gorgeous example of the one central thing you learn when doing empirical economics: Correlation is not causation. Always look for the reverse possibility, or that the two things correlated are both outcomes of something else, and changing A will not affect B.   We seldom get an example that is so beautifully clear.

Update:  Melissa Kearney writes,
"Bill Evans and Craig Garthwaite have an important study [AER] showing that expansions of EITC benefits led to improvements in self-reported health status among affected mothers. 
Their paper provides a nice counterpoint to the Swedish lottery study, one that is arguably more relevant to the policy question of whether more income would causally improve the health of low-income individuals in the U.S.
Thanks Melissa for pointing it out. This is interesting, but I'd rather not get in to a dissection of studies here -- just who takes advantage of EITC benefits, how instruments and differences do and don't answer these problems. The main point of my post is not to answer once and for all the question -- how much does showers of money improve people's heath -- but to point out with this forceful example for non-economists the possibility that widely reported correlations - rich people are healthier -- don't automatically mean that money showers raise health.  

23 comments:

  1. Sweden has free health care and lots of other social insurance. It's not surprising in that context that you would not see much of a relationship.

    ReplyDelete
    Replies
    1. You missed the point. (Or I didn't write it clearly enough.) There is a strong relationship -- wealthier people are healthier, even in Sweden. But when the lottery randomly gives people a lot of wealth, those people and their children do not become healthier.

      Delete
    2. There was a similar study based on the Georgia Land lotteries, which found no improvement in the levels of wealth, income, education, or literacy in the offspring of the lottery winners:

      https://econresearch.uchicago.edu/sites/econresearch.uchicago.edu/files/bleakley.pdf

      Genes matter.

      Delete
    3. The correct interpretation would be that giving people *who play the lottery* a lot of wealth does not make them healthier. This is a special group of people (playing the lottery is not randomly assigned).

      On another note, while richer people may be healthier even in Sweden, the degree of this difference is important for extrapolating to the US. Also the level of overall health is important- perhaps in Sweden everyone is so healthy that the extra wealth doesn't do anything.

      Delete
    4. Regarding the comment by Anonymous on June 14th: It could indeed be the case that the effect of a wealth shock is different in the sample of lottery players we (I am one of the co-authors) study compared to the population as a whole. However, as we discuss in the paper, the lotteries we study were very widespread in Sweden with 50% of Swedes participating in one of the lotteries at the height of its popularity in the late 1980's. Consequently, the lottery sample appears similar to the population as a whole in terms of basic socio-economic characteristics. For these reasons we believe the external validity of our study to be high (albeit not perfect). /Erik Lindqvist

      Delete
    5. It is indeed correct that the effect of a wealth shock may differ in the lottery samples we (I am one of the co-authors) study compared to the population as a whole. However, we believe there are two reasons for why this difference should be quite limited. First, the lotteries we study were very popular in Sweden with 50% of Swedes participating in one of the lotteries at the height of its popularity in the late 1980's. Second, the lottery appear similar to the population as a whole in terms of basic socio-economic characteristics. /Erik Lindqvist

      Delete
  2. The government decides to try to increase the middle class by subsidizing things that middle class people have: If middle-class people go to college and own homes, then surely if more people go to college and own homes, we’ll have more middle-class people. But home ownership and college aren’t causes of middle-class status, they’re markers for possessing the kinds of traits — self-discipline, the ability to defer gratification, etc. — that let you enter, and stay, in the middle class. Subsidizing the markers doesn’t produce the traits; if anything, it undermines them.

    http://pajamasmedia.com/instapundit/106691/
    Dubbed Reynolds' Law
    http://philoofalexandria.wordpress.com/2010/09/25/reynolds-law/

    ReplyDelete
  3. You might add that there is also no strong evidence to suggest that access to health insurance leads to better health outcomes.

    ReplyDelete
    Replies
    1. Whether it leads to better outcomes or not, doesn't it internalize costs in cases where folks without insurance otherwise force healthcare providers to foot the bill/pass on the costs to others?

      Delete
    2. I don't see that as a good thing.

      Delete
  4. It seems like mortality outcomes are a function of all past health behaviors -- the integral over an extended period rather than of behavior at a point in time. In that case it is the relatively high persistence of wealth and the resulting health behaviors and opportunities that transforms wealth into better health outcomes. In that case a one-time windfall, even if it significantly boosts health behaviors/opportunities for a year or two, would not necessarily be enough to outweigh the prior or subsequent years of poverty/low-wealth-induced bad health and generate a significant boost to 10 year mortality.

    ReplyDelete
    Replies
    1. The wealth shocks that drive the identification in our study (I am one of the co-authors) are large also from a life-cycle point of view. Also, it seems to be the case that lottery wins spend their wins over an extended period of time.

      Delete
  5. It seems like mortality outcomes are a function of all past health behaviors -- the integral over an extended period rather than of behavior at a point in time. In that case it is the relatively high persistence of wealth and the resulting health behaviors and opportunities that transforms wealth into better health outcomes. In that case a one-time windfall, even if it significantly boosts health behaviors/opportunities for a year or two, would not necessarily be enough to outweigh the prior or subsequent years of poverty/low-wealth-induced bad health and generate a significant boost to 10 year mortality.

    ReplyDelete
    Replies
    1. According to happiness research, getting a welfare state or winning in a lottery brings only short-lived happiness. Maybe that EITC study also was short-term and self-reported answers mainly reflected happiness, feeling fine. Lottery study probably had a much longer time-span and proved that undeserved money does not cause health even though the winners stayed rich. Health is genes, IQ and self-discipline, which are also pretty much genes. Incomes vary 75% within families and just 25% between different families. It is the individual, not the wealth of their parents.

      Delete
  6. There are other reasons for high taxes, as Obama famously said when told that raising marginal tax rates on the rich does not increase total tax revenues. In policy, symbolism is often more important than real outcomes. Facts matter little, when up against inflexible moral sentiment.

    ReplyDelete
  7. You write

    "These causes can be either "right" or "left" as far as the evidence before us: "Right:" Thrift, hard work, self discipline and clean living lead to health and wealth. "Left:" good parents, good neighborhood, the right social connections lead to health and wealth."

    Aren't these the same? Isn't good parenting in part about putting your kids in good neighbourhoods, teaching them about thrift, hard work, self-discipline etc?

    ReplyDelete
  8. I am not sure why you would expect a sudden positive wealth shock to improve health of an adult. Imagine someone who loves eating ribs, which is unhealthy. More money would simply lead that individual to eat more ribs, or race more motorcycles or drink more champaign, all of which are unhealthy behavior. Now if you took identical twins, separated them at birth and placed them in wealthy vs poor families and track them to adulthood, then I think we could expect the ones in wealthier families to be healthier

    ReplyDelete
  9. Dave Cheppelle already did this study. https://youtu.be/Fg6J1Skptbs

    ReplyDelete
  10. Regarding the comment by Melissa Kearney, one potentially important difference between the study on the EITC by Bill Evans and Craig Garthwaite and our study of Swedish lottery players is the set of outcome variables. It is conceivable that positive wealth shocks have beneficial effects on mental health and stress levels but no (or very small) effects on mortality and serious medical conditions. In fact, we do find that lottery winnings have a small negative effect on consumption of anxiety-reducing drugs and sleeping pills. We tone down this result a bit in the paper (due to the risk of false positives), but it is at least qualitatively consistent with one of the main finds in the paper by Evans and Garthwaite. /Erik Lindqvist

    ReplyDelete
  11. The same genes that drive IQ drive health. IQ is key for wealth. http://www.nature.com/mp/journal/vaop/ncurrent/full/mp2015225a.html

    ReplyDelete
  12. They say money can't buy everything still it provides access to the premium-quality medical services as well as opens a wide range of opportunities for the personal development. I've checked Powerball results USA and discovered that the winners predominantly go broke and end up in misery. Sad but true. Thanks for sharing the info.

    ReplyDelete
  13. Health is more about lifestyle and good values,and yes, a "money shower" does not solve the problem.

    ReplyDelete
  14. Don't we already know that health causes wealth, ceteris paribus? For a start, for any level of talent, ill-health will make you less productive.

    ReplyDelete

Comments are welcome. Keep it short, polite, and on topic.

Thanks to a few abusers I am now moderating comments. I welcome thoughtful disagreement. I will block comments with insulting or abusive language. I'm also blocking totally inane comments. Try to make some sense. I am much more likely to allow critical comments if you have the honesty and courage to use your real name.