The Growing Gap in Life Expectancy by Income Implications for Federal Programs and Policy Responses (2015) / Chapter Skim
Currently Skimming:
3 Growing Heterogeneity of the U.S. Population in Income and Life Expectancy
3 Growing Heterogeneity of the U.S. Population in Income and Life Expectancy
Pages 37-64
The Chapter Skim interface presents what we've algorithmically identified as the most significant single chunk of text within every page in the chapter.
Select key terms on the right to highlight them within pages of the chapter.
From page 37... ...
According to the Congressional Budget Office (CBO) , the share of pretax aggregate household income accruing to households in the bottom quintile of the income distribution fell from 6.2 percent in 1979 to 5.1 percent in 2010 (Congressional Budget Office, 2013b)
|
From page 38... ...
. Despite this narrowing, many studies have found that mortality differences by educational attainment have widened, and likewise by position in the earnings distribution.
|
From page 39... ...
, as well as access to health care, cognitive functioning, and the development of social and psychological resources to seek and preserve health, are the major sources of the mortality differentials. Health behaviors are estimated to account for about 30 percent of the mortality difference for individuals with high versus low levels of education (see summary of the causes for the relation between educational achievement and mortality in Hummer and Hernandez, 2013)
|
From page 40... ...
Many analysts have found that mortality differentials by educational attainment have been widening in recent decades. One prominent study
|
From page 41... ...
When analyzing this measure in relation to mortality, the researchers find no decrease in life expectancy at low educational levels, but they do still find an increase in the mortality gradient over time: However, consistent with other findings (e.g., Waldron 2007) we do see clear evidence for increasing dispersion of survival probabilities between those in the bottom and top of the educational distribution.
|
From page 42... ...
. Recent research has sought to avoid these problems by constructing long-term earnings measures using Social Security earnings histories.
|
From page 43... ...
The bottom half of the earnings distribution was estimated to gain 1.9 years of life expectancy between the 1912 and 1941 birth cohorts, while the top half was estimated to gain 6.5 years of life. Because the Social Security data have limited SES information, Bosworth and Burke (2014)
|
From page 44... ...
. She estimated the increase in life expectancy at age 65 between the top and bottom half of the career earnings distribution of men for the 1912 and 1940 birth cohorts as 5 and 1.3 years respectively." If the data in the middle panel of Table 5 in Bosworth and Burke are used to calculate life expectancy for the top and bottom five deciles for birth years 1920 and 1940, in order to compare to Waldron's estimates, for males the increases were 4.84 and 2.96 years, respectively.
|
From page 45... ...
In summary, an abundance of research over the past two decades finds that SES differentials in mortality are widening, whether SES is measured by educational attainment or income quantile, by composite indices of SES at the county level, or by any of several long-term earnings measures based on Social Security earnings histories. For the purposes of this report, the estimates using career earnings are the most relevant for analyzing the progressivity of government programs and the differential impacts by income class of a menu of possible policy changes.
|
From page 46... ...
The committee experimented with other specifications, including estimating a different mortality factor for each income quintile for each 10-year birth cohort. We settled on the specification just described, which is closer to the Bosworth and Burke specification, because it is less demanding of the limited size of the dataset.
|
From page 47... ...
For the 1960 birth cohort, mortality is not observed at all, because this cohort turns 50 in 2010, 2 years after the 2008 HRS. The age range and number of deaths observed for each of the HRS birth cohorts are illustrated in Figure 3-1.
|
From page 48... ...
Some individuals were dropped from the analysis because of missing values, nonresponse, and similar reasons. SOURCE: Committee generated from Health and Retirement Study data.
|
From page 49... ...
In order to investigate the consequences of mortality inequality for the 1960 birth cohort, we construct a plausible scenario for mortality at ages 50 and above based on our fitted model, on the assumption that the base period trends we observe continue, both in average mortality and, more importantly, in the widening of the differentials by midcareer earnings. This scenario will be referred to as the mortality regime of the 1960 birth cohort, but it is important to keep in mind that it is entirely extrapolated or projected rather than observed.
|
From page 50... ...
. In the baseline analysis, this differential grows from 5.1 years to 12.7 years between the birth cohorts of 1930 and 1960, or by 7.6 years.
|
From page 51... ...
Much more speculatively, we also calculate for some purposes the projected life expectancy at age 50 for the 1990 birth cohort by income quintile, on the assumptions that the underlying trend in mortality decline remains unchanged from the base period and that the underlying trend in differentials about that trend likewise continues unchanged. This second assumption, about the trend in differentials, is particularly problematic because the differentials are strongly influenced by trends in smoking behavior, including both uptake and quitting -- trends that are not expected to continue for long.
|
From page 52... ...
Estimates of mortality differences by income for females are often unstable or present other problems (Waldron,
|
From page 53... ...
The implications for receipt of retirement and health care benefits are clear. But for the 1960 birth cohort, the corresponding probabilities are 66 percent and 26 percent, rising substantially for the top quintile but holding steady or declining slightly for the bottom quintile male.8 The corresponding percentage probabilities of survival from age 50 to 85 for females are 60 versus 46 percent for the 1930 birth cohort, and 77 versus 32 percent for the 1960 birth cohort.
|
From page 54... ...
quintiles and survival might be that trends in income distribution mean that those in the bottom quintile are now poorer relative to those at the top than was the case in the past, and therefore their survival has grown relatively worse. None of the studies just mentioned has addressed this important question, which would require different methods and models.
|
From page 55... ...
GROWING HETEROGENEITY OF THE U.S. POPULATION 55 Survival to 85 -- Females 0.77 0.60 0.51 0.49 0.46 0.47 0.47 0.43 0.37 0.32 1930 cohort 1960 cohort R02856 Survival to 100 -- Females Fig 3-3 panel 3.eps 0.29 0.13 0.06 0.07 0.06 0.07 0.06 0.05 0.02 0.03 1930 cohort 1960 cohort FIGURE 3-3 Continued R02856 Fig 3-3 panel 4.eps the famous Whitehall Studies (see, for example, Marmot et al., 1978, 1991; see Box 3-2)
|
From page 56... ...
Direct comparisons of the CBO projections to the estimates in this report are challenging for several reasons, including the differences in birth cohorts (this report focuses on those born in 1930 and 1960, whereas CBO projections show those born in 1949 and 1974) ; the treatment of those who have qualified for disability insurance (this report includes them, whereas CBO projections exclude them)
|
From page 57... ...
The CBO projections, by con trast, assume that the gradient steepens by 1 to 1.3 months per year, on average, between the 1949 and 1974 birth cohorts. The CBO projections also suggest that the gap increases less for females than males; our projections show the opposite.
|
From page 58... ...
As mentioned before, the influence of smoking and obesity on differentials in mortality depends on how unequally they are distributed in the population. Although tobacco smoking has declined overall, the difference in prevalence of smoking by income level (measured relative to the poverty line)
|
From page 59... ...
Recent analyses (Anderson and McIvor, 2013; Conference Board of Canada, 2014) have documented rising income inequality during 1990-2010; by 2010 the top income quintile received 39 percent of total national income compared with 7 percent for the lowest quintile.
|
From page 60... ...
Body mass index (BMI) equals weight in kilograms divided by height in meters squared.
|
From page 61... ...
Estimates are for current cigarette smoking, age-adjusted to the year 2000 standard population using five age groups: 18-24 years, 25-34 years, 35-44 years, 45-64 years, and 65 years and older. Povertylevel data are percentages of the estimated poverty thresholds set by the U.S.
|
From page 62... ...
Body mass index (BMI) equals weight in kilograms divided by height in meters squared.
|
From page 63... ...
Another important influence on future patterns of mortality differentials will be the impact of ongoing health care reforms in the United States. By providing increased access to health care to groups that lacked access previously, will these reforms produce narrowing gaps in risks of dying across socioeconomic groups?
|
From page 64... ...
We, like others, have used the average of earnings during years when earnings were positive, for a span of 10 years near typically peak earnings. The whole earnings history cannot be used because many workers in earlier cohorts joined Social Security well after they began working, so Social Security earnings histories miss a segment of their earlier earnings.
|
Key Terms
This material may be derived from roughly machine-read images, and so is provided only to facilitate research.
More
information on Chapter Skim is available.