Life expectancy puzzles

For the past 100 years, life expectancy in the US has been on a relentless upward march.  It was 69 when I was born in 1955.  In 1994 it was 75.7.  In 1998 it was 76.7.  In 2002 it was 77.0. In 2006 it was 77.8.  In 2010 it was 78.7.  That’s when Obamacare was passed.  In 2014 it was just 78.8.  That’s the smallest 4-year increase since annual data began (in 1970).

I actually don’t think that’s a very good argument against Obamacare.  But supporters of Obamacare often used arguments that were equally lame, such as the claim that life expectancy in the US is lower than in Western Europe.  Yes, but what does that mean?

To give you an idea of just how misleading life expectancy data can be, consider the case of Hawaii.  This source suggests that Asians in Hawaii live shorter lives than Asians in any other state, in some cases by a wide margin.  (Note Asian life expectancy data is only available for 28 states.)  OK, but how about Hispanics?  Again, life expectancy for Hispanics is lowest in Hawaii, of all the states with data (again 28 states.)  There is no data for blacks in Hawaii.  By now you must have concluded that Hawaii is some sort of hellhole with horrible life expectancies.  And controlling for race it is.  But Hawaii is also the state with the highest overall life expectancy in the entire US.

How is that even possible?  One answer is that whites do fairly well in Hawaii.  But whites are a minority in that tropical paradise.  The real reason is even more bizarre.  Both Hispanics and Asians have inexplicably long life expectancies in the US.

Let’s go back to Obamacare, which was going to improve health by eliminating the problem of people with without health insurance.  The group that is far and away the least likely to have health insurance is Hispanics.  They also have an obesity problem. And yet their life expectancy is 81.8, compared to 79.0 for whites.  Why?  I have no idea. But again, it suggests that health insurance isn’t the issue, something else is going on.

Asians are the biggest group in Hawaii, and they have a mind-boggling 86.5 life expectancy in the US, which is higher than even the richest countries in East Asia (Japan tops out at 84).  And yet Asians in America have a higher poverty rate than whites.  Even if it’s genetic, that doesn’t explain the comparison with East Asia.  Nor does diet.

Once again the state level data might help.  The two longest-lived Asian groups are in New Jersey and Massachusetts, both over 89 years. And it’s pretty clear what’s special about these two states—they both have a lot of recent immigrants who are highly educated and work in high tech (like my wife, who is from China, works in biotech, and will basically live forever, assuming cryonics is perfected within 40 years.)  It’s not surprising that the well-educated live longer, that’s also true of whites.  But nationally there are also plenty of Asians that don’t work in high-tech, so the overall life expectancy is still pretty hard to explain.

So Hawaii is the absolutely worst place for Asians and Hispanics.  But there are so many Asians in Hawaii, and their life expectancy is so mindbogglingly long, that it pushes Hawaii to number one among US states, the very highest life expectancy.  A good example of the need for control variables.

African-Americans present another puzzle.  The black/white gap was 6 years in the early 1980s.  By the late 1980s it had widened to 7 years, and was still 7 years in 1994. No reason for optimism, right?  But then it started shrinking steadily, and was down to 3.8 years by 2010.  As we know, total life expectancy in the US then leveled off, as did white life expectancy.  But black life expectancy kept rising, and the gap is now (in 2014) only 3.4, less than half the gap of 20 years ago.  The gap is still shrinking. Why? I have no idea.  The falling murder rate is one factor, but hardly seems important enough to explain the whole story.  Meanwhile the poor/rich gap is widening among whites, and blacks have some of the same socio-economic problems as poor whites. (Here’s where the Oxy/heroin epidemic may play a role for whites.)

Denmark has the shortest life expectancy in Western Europe.  Diet?  Lots of Danes who moved to the US became Mormons, who adopted their strong communitarian culture.  So Utah has good social indicators, just like Denmark.  Except life expectancy. Whereas Denmark is unusually low for Western Europe, Utah is relatively high for the US.  Another mystery.

My conclusion?  We don’t understand life expectancy very well.  (Commenters: Before giving me your “theory” of Hispanics or Asians, ask yourself if you expected these numbers, before seeing them.)

PS.  After I wrote this I noticed that Alex Tabarrok has a new post on this topic.  His data set stops at 2010, and hence he doesn’t pick-up the leveling off of life expectancy after Obamacare was passed.

PPS.  There are different sources for international life expectancy data, and they do not all agree.


Tags:

 
 
 

44 Responses to “Life expectancy puzzles”

  1. Gravatar of FS FS
    10. May 2016 at 05:13

    Did you see Raj Chetty’s paper last month in the Journal of the AMA? (http://jama.jamanetwork.com/article.aspx?articleid=2513561). Many of the points you discuss here are investigated in far greater detail.

  2. Gravatar of David Pinto David Pinto
    10. May 2016 at 05:26

    I remember reading many years ago that Europe does not count infants who die within 24 hours of birth as live births, but the US does. So Europeans remove some outliers that raise their overall live expectancy compared to the US.

  3. Gravatar of ssumner ssumner
    10. May 2016 at 05:42

    David, How much does that bias the results?

    My general view is that life expectancy doesn’t mean what many people think it means. But I have no idea what it actually does mean. None of the theories I’ve seen match the data.

    On thing is clear. People should not be pointing to life expectancy data in public policy arguments, because there’s no good evidence that public policy is in any way related to life expectancy.

  4. Gravatar of ssumner ssumner
    10. May 2016 at 05:46

    FS, I notice that’s gated. Is there anything on the ethnic differences?

  5. Gravatar of Miguel Madeira Miguel Madeira
    10. May 2016 at 05:48

    “I remember reading many years ago that Europe does not count infants who die within 24 hours of birth as live births”

    This is a myth – at least in Portugal, every child who is born living, is counted as a live birth

  6. Gravatar of asdf asdf
    10. May 2016 at 06:23

    I agree whole heartedly with this post. Life expectancy has minimal statistical value, and close to zero public policy value. When you get right down to it the vast majority of people outside the third world are living to relatively old age, that doesn’t tell you anything useful. Life expectancy doesn’t even have a good correlation with things we would want to measure. Unless your doing related actuarial calculations its a branch of statistics you could probably ignore.

    That’s before even getting into values issues (what trade off is an extra year of life expectancy worth). It’s not even good as a pure statistic.

  7. Gravatar of David Pinto David Pinto
    10. May 2016 at 07:04

    Scott, I do not know how much the difference in counting live births makes overall. I suspect that medical advances are reducing the difference, as more and more children survive the first 24 hours.

  8. Gravatar of Neil S Neil S
    10. May 2016 at 07:19

    Scott,

    Are there any relatively narrow demographic groups (e.g. Swedes, Danes, Japanese, Nigerians, Somalis…) which do not show improved life expectancy in the US versus their country of origin? I am not aware of any.

    Regards,
    Neil

  9. Gravatar of ssumner ssumner
    10. May 2016 at 07:33

    asdf, Good point.

    Neil, Good question.

  10. Gravatar of HW HW
    10. May 2016 at 07:51

    Now this is an interesting topic.

    I would like to note that life expectancy for blacks in England (not the UK!) is probably higher than for non-Hispanic whites in the US, based on a 2001 report where blacks lagged the the English average by only 0.5 years. Meanwhile, life expectancy in the US lags England by about 2 years today, so blacks in England probably live 1.5 years longer than whites in the US.

    Also, in Sweden the life expectancy of Chinese women lags the national average by 1.1 years! Meanwhile, the life expectancy for Americans in Sweden is higher than the national average. Data from Statistics Sweden.

    So, the ambiguity and lack of patterns also exists in other countries.

    Sources.
    http://www.cpa.org.uk/information/reviews/thefutureageingoftheethnicminoritypopulationofenglandandwales.pdf

    http://www.scb.se/Grupp/Valfard/BE0801_2004K03_TI_09_A05ST0403.pdf (Swedish)

  11. Gravatar of HW HW
    10. May 2016 at 08:05

    I forgot to mention, Chinese people living in England also have a lower life expectancy than the average English, according to that report. Weird.

  12. Gravatar of Justin Irving Justin Irving
    10. May 2016 at 08:10

    Anyone who reads Razib Khan or Greg Cochran knows the answer to this. There are all sorts of biochemical differences between population groups, some medications for example don’t work, or even have negative effects when used on certain population groups. It’s likely that East Asians simply have a longer life expectancy, the way a King Charles Cavalier lives longer than a Great Dane. Latin American Mestizos, who have significant deep ancestry overlap w/ East Asians also live longer than Europeans, despite worse access to medical care, more dangerous jobs and lower incomes.

  13. Gravatar of Airman Spry Shark Airman Spry Shark
    10. May 2016 at 08:52

    “Let’s go back to Obamacare, which was going to improve health by eliminating the problem of people with health insurance.”

    Freudian slip?

  14. Gravatar of Steve Steve
    10. May 2016 at 10:41

    Obamacare, at the margin, probably expanded access to opiates, and could be lowering life expectancy as a result.

  15. Gravatar of Steve Steve
    10. May 2016 at 10:53

    Also, how is life expectancy calculated? Average death age in a particular year?

    We could be starting to see the lagged health effects of 2008-09 job loss showing up in the 2010-14 life expectancy data.

    Obamacare only affects 10% or so of the population, so it shouldn’t have much effect on life expectancy. It would be interesting to see if the life expectancy of small business owners is dropping, though.

  16. Gravatar of Doug M Doug M
    10. May 2016 at 11:19

    The biggest influence in life expectancy is infant mortality. One child that dies as a baby, balances out with a half dozen others who live an additional 10 years.

    As our healthcare improves, there are some reasons to think that infant mortality will start to increase. More babies that would have been considered non-viable at birth, now go to neo-natal intensive care and survive for a short amount of time

  17. Gravatar of Doug M Doug M
    10. May 2016 at 11:20

    Another factor to consider are those who die because of accidents.

    A small increase in auto safety may have a large impact on life expectancy.

  18. Gravatar of Floccina Floccina
    10. May 2016 at 12:27

    One reason Hispanics and Asian live longer. Lower rates of twinning. http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.442.7724&rep=rep1&type=pdf

  19. Gravatar of PG PG
    10. May 2016 at 12:49

    Here’s a theory:

    “Working hard” is the confounding variable. Women, Asians, Hispanics, and high-income individuals tend to work hard. People who work hard are more satisfied with their lives, and live longer.

  20. Gravatar of Floccina Floccina
    10. May 2016 at 12:55

    Denmark has the shortest life expectancy in Western Europe.

    Greenland is part of Denmark and the Inuit do not do so well.

    https://en.wikipedia.org/wiki/Demographics_of_Greenland#Life_expectancy_at_birth

    I wonder if the Inuit population is enough to have a significant effect in overall life expectancy.

    Italy does very well in life expectancy compared does very well compared to the northern Europeans see: https://en.wikipedia.org/wiki/List_of_countries_by_life_expectancy

    Could that be because Italy does has fewer child maltreatment deaths see:
    http://theincidentaleconomist.com/wordpress/maltreatment-of-girls-and-adolescent-pregnancy-in-the-us/

    I would think that is because Italy has less a problem with alcohol.

    I second Steve’s statement:
    Obamacare, at the margin, probably expanded access to opiates, and could be lowering life expectancy as a result.

    And add Medicare part D see: http://marginalrevolution.com/marginalrevolution/2016/05/demand-curves-slope-downward-opiod-edition.html

  21. Gravatar of ssumner ssumner
    10. May 2016 at 15:08

    HW, Very interesting.

    Justin, Yes, we know that East Asians have a longer life expectancy, but why do Asian Americans live longer than East Asians?

    Airman, Oops

    Steve, I think it’s the age at which 50% of people would be dead, calculated using death rates for each age. But I’m not certain.

    Steve and Doug, I seem to recall that recessions improve health.

  22. Gravatar of ssumner ssumner
    10. May 2016 at 15:09

    Floccina, Interesting. I think Greenland’s population is too small to make much difference.

  23. Gravatar of HW HW
    10. May 2016 at 17:24

    I’m not sure why people take it for granted that East Asians live longer than whites. Certainly this isn’t the case globally. China makes up more than 80% of the population of East Asia and has a lower life expectancy than many if not most white-majority countries. Among the developed countries, South Korea and Taiwan don’t really stick out either compared to other OECD countries. Hong Kong and Singapore do better, though they are still even with Southern European countries like Spain and Italy (plus the life expectancy tend to be longer in cities). It’s really only Japan that’s the outlier in East Asia, whose population makes up less than 10% of East Asia. And even in Japan it’s only 1 longer higher than in Spain or Italy.

    It’s true that Asian-Americans (not just East Asians though) have a very long life expectancy in the US, but so do Latinos, despite Latin American countries not really sticking out on that front. The same goes for foreign-born blacks in the US, whose life expectancy is 2 years longer than that of Japan at age 65! So the higher life expectancy seems to be a trend among foreign-born Americans in general, and not just Asians. And yes, most Asian-Americans are foreign-born (about 74% according to a Pew Report: http://www.pewsocialtrends.org/2012/06/19/the-rise-of-asian-americans/).

    And as I’ve already mentioned in a post above, the higher life expectancy among foreign-born East Asians isn’t observed in some other Western countries where I could get the data. In England and Sweden, those of Chinese origin don’t have higher life expectancy than the country averages, or compared to immigrants from other countries. In fact I can see no racial pattern from the data.

    Those who are arguing that East Asians have a higher life expectancy thanks to “biochemical differences” need to first establish that East Asians have a higher life expectancy at all. Japan is only one country and Asian immigrants in the US share this trait with Latinos and foreign-born blacks.

    Study on life expectancy among foreign-blacks in the US: http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0037177

  24. Gravatar of HW HW
    10. May 2016 at 17:56

    To clarify, foreign-born blacks live longer than foreign-born whites in the US, and a lot longer than native-born whites.

  25. Gravatar of dtoh dtoh
    10. May 2016 at 22:15

    Scott,
    You need to normalize for
    – Infant mortality (different countries use different metrics)
    – Auto fatalities
    – Homicides

    That will then tell you something about health and health care.

    The normalize for obesity and it will tell you something about medical care.

    The raw life expectancy numbers are virtually meaningless and not worth blogging about.

  26. Gravatar of colllin colllin
    11. May 2016 at 07:23

    Of course life expectancy is extremely tough to measure as it measures variables that last a lifetime. And better yet the measuring minority backgrounds in individual divides the data incredible amounts. (Reviewing the file suggest living in Hawawii and warm climates is the best way to extend life expectancy. Where is Alaska?) Other points:

    1) I suspect one reason why the gap of white and minority life expectancy is shrinking is urban population live longer. Could be as simple as emergency healthcare is quicker? Minority variables is either noise or minor variables.
    2) Why did life expectancy slow down after 2010? Although it is counter to most thinking, but don’t early recessions increase life expectancy for a short period. (Lower levels of stress and less driving likely here.) So after 2010, life expectancy was expected to slowdown with the increase 2009.

  27. Gravatar of Michael Larsen Michael Larsen
    11. May 2016 at 12:01

    The relatively low danish life expectancy can be explained by many female smokers and a high average alcohol consumption. Greenland only accounts for 1% of the total danish population and is not part of the statics for southern Denmark. An often used phrase in Denmark is “does not apply for the Faroe Islands and Greenland”.

  28. Gravatar of Lorenzo from Oz Lorenzo from Oz
    11. May 2016 at 13:05

    As far as I am aware, every immigrant group becomes richer and longer lived in the US. (Swedish Americans are richer and have a lower poverty rate than Swedes, for example.)

    Probably has something to do with why so many folk come to the US: it is rational revealed preference. (Provided, of course, you have a sufficiently low attachment to living immersed in your original culture.)

  29. Gravatar of ssumner ssumner
    11. May 2016 at 16:33

    HW, Good comment, but I’m not quite convinced. South Korea’s life expectancy is rising very rapidly, and is now 9th in the world. I predict it will soon be near the top. Recall that many South Koreas were young during the lean years, when Korea was extremely poor. Over time that group will wash out of the data.

    I’m also not entirely convinced that you can throw out places like Singapore and Hong Kong. Let’s say it’s still an open question.

    The US Asian data is rather astounding, and hard to explain away.

    dtoh, You said,

    “That will then tell you something about health and health care.”

    Those adjustments would help, but even then I doubt it would tell you anything about health care.

    Collin, I heard that about warm climates as well, but it’s odd that Asians in Hawaii live shorter lives than any other state.

    Michael, Thanks for that info. I think fatty foods may be an issue as well.

    Lorenzo, I’ve heard that as well. When I think of all the ways America is screwed up, I can’t even imagine what other countries are doing to end up even poorer than us.

  30. Gravatar of HW HW
    11. May 2016 at 17:41

    I think the evidence is too limited to support the premise that East Asians have the longest life expectancy (for now).

    As far as the US goes, basically all immigrants have a longer life expectancy than native-born (I’m not saying all US Asians are immigrants, but 74% are foreign-born, which skews the data). Foreign-born blacks live 2 years longer than people in Japan using one methodology. The fact that Latinos still have a lower life expectancy than US Asians (though still 3 years longer than whites) could be explained by the fact their median household income is only like 50% that of US Asians. So it’s possible the “healthy immigrant effect” plays a large role in this.

    As for Hawaii, most US Asians there aren’t immigrants; they have lived there for generations. Thus they don’t benefit from the “healthy immigrant effect” and their life expectancy converges closer to the American mean. Coincidentally, Asians in Hawaii do worse in school than Asians in pretty much all states. In fact whites in Hawaii outperform Asians there (according to NAEP data), another example of convergence to the American mean as descendants of immigrants assimilate.

    It is in any case unreliable to base assumptions about East Asian life expectancy off data on US Asians alone. Data from other countries give a different picture. And it’s true that South Korea and Taiwan were as poor as Sub-Saharan African countries and their life expectancy may yet catch up to Japan’s, but this is all assumptions for now. Except for Japan (less than 10% of East Asia’s population) and US Asians, most East Asians aren’t at the top of the life expectancy table. I’d be careful to draw a conclusion based on those two examples alone, when there is also contrarian data, and would take a more agnostic stance (until we see how things turn out in these countries, as well as China).

    NAEP scores by race and state: http://www.nationsreportcard.gov/reading_math_2015/#mathematics/state/gaps?grade=8

  31. Gravatar of ssumner ssumner
    12. May 2016 at 10:13

    HW, You said:

    “As for Hawaii, most US Asians there aren’t immigrants; they have lived there for generations. Thus they don’t benefit from the “healthy immigrant effect” and their life expectancy converges closer to the American mean. Coincidentally, Asians in Hawaii do worse in school than Asians in pretty much all states. In fact whites in Hawaii outperform Asians there (according to NAEP data), another example of convergence to the American mean as descendants of immigrants assimilate.”

    That’s a very interesting hypothesis, you might be right. I’ll keep an open mind on this question, although I still lean a bit toward the other view. If South Korea doesn’t move well ahead of Europe in 10 years, I’ll change my mind.

  32. Gravatar of Gimlet Gimlet
    12. May 2016 at 11:01

    “Asian” is such a huge category. I wonder if there are concentrations of particular Asian subgroups in certain states that affect the health outcomes? For instance, in Minnesota, the second lowest-ranked state for Asians, they have a concentration of Hmong people. And for Hawaii, I wonder if they include Native Hawaiians (or other Pacific Islanders like Tongans or Samoans) as Asians? Or people who are half or a quarter Native Hawaiian and the rest Asian?

  33. Gravatar of Floccina Floccina
    13. May 2016 at 09:19

    You are correct Scott, I looked up the numbers and Greenland only brings the average life expectancy of Denmark down about 1/10th of a year.

  34. Gravatar of Quote of the Day | The Locker Room Quote of the Day | The Locker Room
    13. May 2016 at 09:23

    […] From Scott Sumner at TheMoneyIllusion: […]

  35. Gravatar of Jim M. Jim M.
    13. May 2016 at 10:55

    It should be noted that foreign-born Americans of all ethnicities have very low infant mortality.

  36. Gravatar of Careless Careless
    13. May 2016 at 11:13

    East Asians enter puberty and menopause at a later age than others. Sounds like it’s mostly just a longer natural lifespan

  37. Gravatar of liberalarts liberalarts
    13. May 2016 at 18:59

    Is it possible that immigrants often arrive after childbirth, so that their ethnic group has fewer chances of infant mortality (their children already having survived or not survived childbirth elsewhere?

  38. Gravatar of ssumner ssumner
    13. May 2016 at 19:01

    Floccina, Thanks for checking.

  39. Gravatar of J J
    13. May 2016 at 21:27

    The biggest influence in life expectancy is infant mortality. One child that dies as a baby, balances out with a half dozen others who live an additional 10 years.

    Obviously, populations with less births will have less infant deaths and in average will live longer. Increase of life expectancy is linked to decreasing fertility.

  40. Gravatar of I I
    14. May 2016 at 00:24

    For those asking how life expectancy is calculated, it is the mathematical expectation of age at death using probablity of death at each individual age.

    The probabilities can be calculated in two different ways:

    1) period life expectancy: the conditional probability of dying at age x having reached age x is the actual experienced death rate in the population for people aged x in, say, the last year

    2)cohort life expectancy: same as above but an assumption is made for death rates falling in future years, ie it assumes that actual life expectancy of people living now would be understated by (1).

    Period life expectancy is the statistical measure and is pretty robustfor large populations. Cohort life expectancy requires “expert judgement” but is needed for policy making, pensions, insurance, etc

    Life expectancy isn’t affected by the underlying age distribution of the population or the number of births. It also isn’t a median, it is a mean.

    Cohort life expectancy doesn’t give sensible comparisons between countries because it is based on differing assumptions. Period LE does and I think that is what’s used in the post.

    People looking at mortality patterns will generally try to split populations into relatively homogeneous subgroups for analysis, so that shifts in the make up of a population cam be allowed for in models.

  41. Gravatar of Art Deco Art Deco
    14. May 2016 at 09:36

    In fact whites in Hawaii outperform Asians there (according to NAEP data), another example of convergence to the American mean as descendants of immigrants assimilate.”

    Orientals in Hawaii are locals with an unremarkable occupational distribution. Most residents of Hawaii are oriental, so the sort of Jewish occupational distribution (heavy on the professional-managerial bourgeoisie with few wage-earners and almost no lumpenproletariat) you see on the mainland among ethnic Chinese and ethnic Japanese (though not, perhaps, the Filipinos and Polynesians common in Hawaii and unusual elsewhere). Essentially, Japanese and Chinese in Hawaii are people too attached to their home to go outside Hawaii to study and work or are ordinary people for whom professional-managerial jobs would not be a realistic aspiration in any set of circumstances. Hawaii Japanese have a unique sensibility many generations in the making. Haolies in Hawaii include some people with a certain amount of history on the islands – and a huge mass of uprooted souls who went there for the climate and put up with the housing prices. You don’t go and live in Honolulu if you want to live in a place with texture. It’s ticky tacky.

  42. Gravatar of Carsten Carsten
    14. May 2016 at 13:38

    Being a Dane, I am surprised to see the claim that ´many Danes became Mormones´ this is new to me, and I think a myth, so please substantiate. Low Danish life expectancy is clearly related to unusually high alcohol consumption, and a pretty fat diet. Also Danish health care have clearly been deteriorating…

  43. Gravatar of DocMerlin DocMerlin
    16. May 2016 at 05:51

    Low birth rate doesn’t explain why Latinos have such high life expectancy rates, as Latinos have a high birth rate.

  44. Gravatar of I I
    17. May 2016 at 21:56

    Birth rate has no impact on life expectancy. LE does not depend on the age distribution of the population, only on observed death rates at individual ages.

Leave a Reply