• Subtleties of life expectancy – ctd.

    Debate over Alan Simpson’s comments on life expectancy continues.

    The last chart from my previous post on this topic showed differences in life expectancy by race. What you’d really like, however, is differences by socioeconomic status. After all, it’s far more likely that we can (and perhaps should) base policies on earnings rather than race. Unfortunately, the CDC data I used two days ago didn’t have differences by earnings.

    But then I received an email from Paul Van de Water, pointing me to a paper by Hilary Waldron that appeared in Social Security Bulletin in 2007. It’s entitled, “Trends in Mortality Differentials and Life Expectancy for Male Social Security–Covered Workers, by Socioeconomic Status.” She did the work for me.

    Let’s start with a chart I made from her paper (Table 4):


    What you’re looking at is the life expectancy of a male who reached age 65 in 1977-2007. The blue line is the top 50% of earners; the green line is the bottom 50%. While the top half of earners have seen an increase of their life expectancy at 65 rise about 5 years over these three decades, the bottom half saw their life expectancy at 65 rise barely a year.

    Think about that when advocating for an increase in the age of eligibility because “everyone” has seen their life expectancy increase.

    Here’s another chart I made from Hilary Waldren’s data (Table 9) comparing the life expectancy at age 65 for Social Security-covered male workers in four different earning percentile brackets in the US and other OECD countries:


    While the richest 25% of Americans compare favorably to any other country, those in the bottom 50% of earners don’t. In fact, half our country can expect to live shorter lives from age 65 than the average person in almost all of these countries.

    Think about that when arguing that the richest country in the world can’t afford Social Security and Medicare starting at age 65 because “everyone” in this country has seen their life expectancy increase.

    Update: Added in US Total for All Social Security Covered workers.  Note that this does not include American males who are not covered by Social Security.

    • Aaron: I’m not clear about which position your last sentence would take (“Think about that when arguing that the richest country in the world can’t afford Social Security and Medicare starting at age 65 because “everyone” in this country has seen their life expectancy increase.”)

      For instance, raising the minimum age at which we can claim SocSec would affect a greater portion of the top quartile of Americans; those folks should then have a greater stake in keeping that minimum age where it is today, while you indicate that it is the poorest that have the most to lose.

      Moreover, another comparison that should probably be made with the second chart above (showing the US quartiles) with the same chart from the year 1960 or 1970, similar to the one you show in your original post. Perhaps the lowest quartile is out-trending the highest? Would our story change in that case?

      • I hope you are not a credentialed actuary. It would be bad for the profession if one of its own wrote that “raising the minimum age at which we can claim SocSec would affect a greater portion of the top quartile of Americans”. High earners live longer and are less likely to retire at 62, so fewer of them would be affected.

    • So your data shows that people who earn more money and can afford a more comfortable retirement and better overall care tends to live longer??? That’s SHOCKING!

      • A deeper read might be that health affects income and longevity. Causality is not at all clear in this picture. Yet it reveals an important fact as we think about Social Security, not to mention health care.

    • Why does a number make a difference? Every 80-something I’ve ever met did not look very happy at reaching their age.

    • The difference between in life expectancy between those in the top 50% and bottom 50% is truly shocking. Interesting that countries that tend to have lower income inequality seem to overall do much better than those with higher inequality. It’s crazy that the average for all of both Japan and Iceland is still higher than the top 25% of Americans.

      Given that income inequality has grown since 1980 I wonder what the impact will be going forward?

      • Japanese Americans outlive Japanese in Japan. Hispanic Americans out live non-Hispanic Americans. (And BTW Hispanic Americans have lower infant mortality than non-Hispanic Americans.)

        So do not worry correlation in this case dose not imply causation.

      • Also:


        Indeed, the health-income gradient is slightly steeper in Canada than it is in the U.S.

        Does Canada’s publicly funded, single payer health care system deliver better health outcomes and distribute health resources more equitably than the multi-payer heavily private U.S. system? We show that the efficacy of health care systems cannot be usefully evaluated by comparisons of infant mortality and life expectancy. We analyze several alternative measures of health status using JCUSH (The Joint Canada/U.S. Survey of Health) and other surveys. We find a somewhat higher incidence of chronic health conditions in the U.S. than in Canada but somewhat greater U.S. access to treatment for these conditions. Moreover, a significantly higher percentage of U.S. women and men are screened for major forms of cancer. Although health status, measured in various ways is similar in both countries, mortality/incidence ratios for various cancers tend to be higher in Canada. The need to ration resources in Canada, where care is delivered “free”, ultimately leads to long waits. In the U.S., costs are more often a source of unmet needs. We also find that Canada has no more abolished the tendency for health status to improve with income than have other countries. Indeed, the health-income gradient is slightly steeper in Canada than it is in the U.S.

      • What percentage of Japan’s population live to be over 65 to start with? It does not say that ALL Japanese have a higher life expectancy than the US.

    • One has to be careful about what implications to draw from these data. I see that Matt Yglesias and Ezra Klein have suggested that differential life expectancy means that raising the retirement age is a regressive benefit cut. Neither of them explicitly compared it to an across-the-board cut, but it is hard to interpret what they wrote as not implying that an increase in NRA is more regressive than a flat-percentage cut. I respect both of them a lot, but this interpretation is incorrect.

      A change in NRA cuts everybody’s benefit roughly equally, with a slightly greater percentage cut for those who retire later (because the delayed-retirement credit is 8% per year, while the early-retirement deduction is 6.67% per year). So, to the extent that higher earners tend to claim benefits later than lower earners, an increase in NRA is (very) slightly more progressive than an across-the-board cut. For all practical purposes, those two types of benefit cuts can be viewed as identical. There are no facts about mortality that can alter this equivalence.

      It is, of course, true that any proportional cut in Social Security benefits will have a larger impact on lower earners because they are much more likely to depend on Social Security for most or all of their retirement income. But again, this is equally true whether the cut is labeled an “across-the-board benefit cut” or an “incerase in the retirement age”. Also, this difference in impact is not what is usually called a regressive benefit cut (even if perhaps it should be – I am sticking to facts here, not opinions).

      Now, a change in the earliest age of eligibility for Social Security benefits (the EEA) would indeed disproportionately affect people with short life expectancy, whose lifetime benefit is maximized by earliest possible retirement. Given the correlation between life expectancy and lifetime earnings, an increase in the EEA is a regressive benefit cut. However, this is not the usual “increase in the retirement age” that is commonly proposed (and currently being phased in as part of the 1983 reform), and there is a good reason it is not so common: it has little impact on the system’s solvency. (It is not a benefit cut at all unless your life expectancy is shorter than average, and most of the cut would not even be recognized by current valuation methods.)

    • The bottom graph is meaningless without comparing against comparable quartiles for other countries, or without showing the aggregate USA line. Just eyeballing the data it would appear that the entire USA would fall in right about where Norway is – a country which incidentally has relatively low income inequality and a highly socialized medicine system. I’m not sure what conclusions, if any, can be drawn from this analysis.

      • Fair enough. I added in the aggregate data.

        You should also be careful about making the perfect the enemy of the good. These data may not be everything you want them to be, but that doesn’t render them “meaningless”.

        • Sooner or later, somebody will link to (or repost) the graph out of context, so it would be helpful to put “at age 65” in the graph title, and not rely on people reading the text around it.

          If I remember correctly, the US-OECD comparison looks better at age 65 than at birth because we, too, have universal health insurance for people over 65, plus our life expectancy at birth is affected by higher infant mortality, traffic accident and homicide rates. (AIDS may still be a non-negligible factor, too.)

    • The comments about an increase in the retirement age being progressive puzzle me. I guess I can kinda see how that could be if you look only at monthly income. But as you increase the retirement age, total income over the expected life clearly is less affected for those with longer life expectancies(rich) than for those with shorter (poor). I.e. regressive. Am I missing something?

      • I assume you are referring to my comment. It is essentially a flat cut, neither progressive nor regressive. I did say “very slightly more progressive” than an across-the-borad cut, but I also carefully explained that the difference was insignificant and it should be viewed as completely equivalent. So describing my comment as calling it “progressive” would be seriously distorting.

        And you are missing something, because your statement about the total income over the expected life is clearly incorrect. I don’t know how you got to that conclusion. If your montlhy benefit is cut by, say, 6%, then your lifetime benefit will also be cut by 6%. It doesn’t matter how long you are going to live and receive benefits, one month or 50 years, your lifetime cut is 6%.

        Perhaps you mean that Social Security benefits are a greater proportion of total lifetime income (i.e., including wages, salaries, other pension, etc.) for the lower earners, so any benefit cut takes away more of a poor person’s total resources than a rich ones. That is true, and in that sense, any flat across-the-board Social Security benefit cut makes the overall lifetime income distribution more regressive. However, then the argument that it is regressive because poor people don’t live as long as rich people becomes even more wrong – the regressivity (in this sense) of a benefit cut would be greater if poor people lived longer than rich people.

    • While the top half of earners have seen an increase of their life expectancy at 65 rise about 5 years over these three decades, the bottom half saw their life expectancy at 65 rise barely a year.

      Not only do they live longer but they get more form SS each month. SS is a very bad program. If voters were rational we might replace it with a weekly check to every adult citizen.

    • What is shown here is that people who are less intelligent have the least earning jobs and die earlier.

      Stupidity => bad earning.
      Stupidity => die early.

      and NOT:
      Bad earning => die early.

      • No causal claim has been made. There are other explanations, including poor health (which can be random and not due to “stupidity) leads to lower earnings (studies show this) and to an earlier death (well, of course).

    • There is always a temptation to assume causality, so reminders to avoid this are welcome. However, the idea that…

      Stupidity => bad earning.
      Stupidity => die early.

      … goes beyond assuming causality and leaps directly into assuming facts without evidence (as well an amazing lack of imagination and/or narrow experience).

    • in western europe men live longer than their counterparts in the united states even though the do not live in “the greatest country in the world’, do not have the benefit of the best health system in the world, and the benefit of the greatest economic opportunities for everyone”

    • so here’s how I see it… canadians and most of the eu live longer, at 30-50% less cost, with everyone insured, lower mortality rates for almost all causes of death, and virtually comparable cure rates… duuuuuuuuuuuuuh !

    • Question ? … How many dismal scientists can you place on the head of a pin ?