The Gran Torino problem

The Case and Deaton paper revealing the worsening health of white working-class Americans.

What’s the matter with middle-aged working class white Americans?

06114132_Clint Eastwood’s Gran Torino hasn’t yet made it into Keith and Johann’s film selections, but no doubt will. It’s a witty and moving, if formulaic, exploration of the psyche of the American white working class, represented by Eastwood’s character Walt Kowalski, a retired car worker in industrial Michigan.

The film has a happy ending. Not so real life. The health of middle-aged white American men with no college education has been deteriorating to a surprising and shocking extent. RBC readers deserve for the weekend a comment thread on this now famous chart:


The article (full text here) is a fine piece of old-fashioned statistical reportage. Perhaps you need a Nobel Prize in Economics, like Angus Deaton (co-author with Anne Case), to get away with an important paper free of equations. They left black men off the chart, I suppose because the line would be above the white men, as mortality is still higher than for whites, obscuring the main point. But it declines, parallel to the American Hispanics and everybody else.

Key data elsewhere in the paper:

  • the rising mortality is concentrated in those without college education;
  • the increase is largely accounted for by “poisonings, suicide, chronic liver disease, and cirrhosis”;
  • the deterioration was paralleled in morbidity, including reported pain and mental health;
  • the geographical distribution is all over the place – it’s not concentrated in the South, and all mega-regions show a wide dispersion between census areas.

Paul Krugman describes the finding as a portrait of “a society gripped by despair”. The nearest historical analogy is the fall in life expectancy of Russian men after the collapse of the Soviet Union, which combined a dramatic loss of economic security and the uprooting of an entire social structure based on a failed ideology. But what peculiar causes drove a change in the USA that wasn’t general, or general for men, but limited to white working-class men?

Inequality increased in other OECD countries, where health continued to improve. Case and Deaton and .. suggest the cause might be an increase in economic insecurity. Didn’t this affect working class blacks and Hispanics just as much? And women?

It is tempting to point to the cultural shocks some of which Eastwood explored in Gran Torino : feminism, immigration, the narrowing of the opportunity gap for blacks and Latinos, and gay rights, in sum a striking loss of relative status compared to other groups of Americans and foreigners. In addition, the collective spokesmen for the white working class – the enfeebled unions, the progressive wing of a Democratic party becoming a rainbow coalition of minorities and women – ceased to be effective. These factors were present for working-class white British and French men too of course: look at New Labour. Add in the economic insecurity and secular wage freeze, present in other OECD countries but to a lesser extent. It’s possible that the decline in unionized mining and manufacturing in the USA hit white workers more than minorities, because unions often accepted or reinforced discrimination by employers. The final newt in the witches’ cauldron was the extreme weakness of the American safety net and patchy health care. If you throw all these together, you get a set of factors specific to white American working-class men.

This can be called the “perfect storm” theory. It takes the Job-like assault on many fronts to produce the flight to drugs, alcohol and suicide. A German man on the assembly line at Volkswagen also had to cope with uppity women, gays and immigrants; but he kept his job and pension. A British miner lost his job, but the NHS was still there for his health care.

It’s an optimistic theory. For if it takes the full conjunction of factors to produce rising death rates, tackling any one will have a large restorative effect. It would be much easier to repair the safety net than to restore the high-wage, unionized manufacturing jobs Erik Loomis pines for.

Question for Daniel Kahneman: are we biased to look for single causes rather than multiple ones? The general confusion over oncogenes and environmental cancer triggers (SFIK you need both) suggests that this may be so.

If I did this for a living, I would dig into the geographical variation. There are sub-regions in all parts of the USA that look like France, others that look like Russia. The same general causes were operative in both, but the perfect storm theory could in principle generate the dispersion. For example, the Bay Area of California is exceptionally full of patriarchy-destroying uppity women, Latinos and gays, and I suppose very unequal in income from the winner-take-all capitalism of Silicon Valley, but a working man in Oakland can get a job and the health care safety net in California is better than elsewhere. What’s the death rate?


Update 8 November

Kevin Drum points out that Case & Deaton also provide a chart showing sharply rising death rates from “poisonings, suicide, and liver” among whites of both genders and all age groups. C&D: “the midlife group is different only in that the sum of these deaths is large enough that the common growth rate changes the direction of all-cause mortality.” This counts against my synergetic hypothesis; the effects may simply be additive. Where is factor analysis when you need it?

Author: James Wimberley

James Wimberley (b. 1946, an Englishman raised in the Channel Islands. three adult children) is a former career international bureaucrat with the Council of Europe in Strasbourg. His main achievements there were the Lisbon Convention on recognition of qualifications and the Kosovo law on school education. He retired in 2006 to a little white house in Andalucia, His first wife Patricia Morris died in 2009 after a long illness. He remarried in 2011. to the former Brazilian TV actress Lu Mendonça. The cat overlords are now three. I suppose I've been invited to join real scholars on the list because my skills, acquired in a decade of technical assistance work in eastern Europe, include being able to ask faux-naïf questions like the exotic Persians and Chinese of eighteenth-century philosophical fiction. So I'm quite comfortable in the role of country-cousin blogger with a European perspective. The other specialised skill I learnt was making toasts with a moral in the course of drunken Caucasian banquets. I'm open to expenses-paid offers to retell Noah the great Armenian and Columbus, the orange, and university reform in Georgia. James Wimberley's occasional publications on the web

13 thoughts on “The Gran Torino problem”

  1. I haven't read the paper (and may not be well equipped to understand it if I did), so it's possible that this is addressed in it. But the kind of loss of status and security that James points to would surely be a far different experience for a white working-class man of 45 in 1990 as compared with a white working-class man of 45 in 2013. The latter might never have possessed the sort of status and security that the former would have felt that he had lost. Wouldn't it be useful to break this out by one-year age cohorts, rather than ten?

    1. The paper is remarkably straightforward. That helps explain its large media impact.

      Your argument is testable as it leads to the prediction that the deterioration will bottom out soon. People get used to anything.

      Your suggestion for a finer-grained analysis by age goes with mine on the geographical breakdown in the wish-list I'll pass on to Case and Deaton, when the comment thread is full. And to Harold, why not.

    1. Your link doesn't work, but I figured out how to get there. Some very interesting stuff, both by Andrew Gelman and his remarkably astute commenters. In the post you attempt to link to, Gelman presents the very analysis by one-year age range that I ask for above. In some of the commentary, it's also suggested that a possibly more useful or revealing breakdown would be by year of birth rather than by age.

  2. "Social" factors of various types should indeed be explored here, and may well be most or all of the driver. But the explosion of prescription painkiller use is also a highly plausible culprit. For lot's of reasons use of opiates has gone up dramatically, and opiates can be seriously bad news in some circumstances.

    1. Absolutely. But why did this disproportionately impact white Americans? The opioids, careless physicians, booze, and guns for suicide were presumably also available to blacks and Hispanics, who also lost secure jobs under the likes of Cary Fiorina, had their homes foreclosed, etc. So you do need sociocultural cofactors.

      1. Actually, there is a substantial racial disparity in prescription of opioids (see the paper at…. This, as well as alcohol and drug abuse in their teens and twenties could account for the mortality increase. Recent comments at Gelman also suggest this trend may be in part an artifact of data collection (so called "bad binning").

      2. Maybe because this is a spike in poisoning/intoxicant deaths among white people, which is only going to be visible against a relatively low previous rate?

  3. I'm a little surprised by the surprise at this. What do you think happens when you take security and hope away from people?

    The fact that people of color and women deal with this lack better — being long used to it – shouldn't exactly be something to brag about, now should it?

    Feel the Bern. Smell the Tea Party. Etc etc.

    1. Recent reanalyses of this data seems to suggest that in fact, the increased death rate is more conspicuous among white women than white men. Why that might be so is beyond me. I believe we're only at the beginning of figuring this whole thing out. One thing I'd look at is sex, as well as race, differences in prescribing opiates, and how they may have changed over time. Also, I wouldn't assume that cultural despair is principally a male thing.

  4. I have not tried to run the data to test this hypothesis, but I wonder how much of this is alcohol-related.

    One key change that drives alcohol use is religion; traditionally, Baptists and Methodists have strongly opposed drinking, and most other churches have not–but Baptists and Methodists were much more working-class than other Protestants. (Catholics were likely to be working-class.)

    I'm wondering if the general loss of religious affiliation among the working class significantly increased alcohol use in the populations that used to be working class Baptist/Methodist.

    This would help explain why the trend is different for Blacks (much less change in religious affiliation) and Hispanics (primarily Catholic, so religious affiliation wouldn't have reduced alcohol use as much.)

Comments are closed.