The Socioeconomic Gradient in U.S. Age-Related Transfers, Health, and Mortality

The goals of this research are to reconstruct the National Transfer Accounts (NTA) for the U.S. by socioeconomic status (SES) at two points in time, and then to assess how changes in these age-related transfers by SES may help to understand trends in health and mortality among older Americans as measured in the U.S. Health and Retirement Study (HRS). Two recent developments emphasize the salience of this investigation: recent declines in U.S. life expectancy at birth overall that appear to be partially attributable to trends among specific subgroups defined by ethnicity and SES, and new research highlighting the very different trends in U.S. mortality according to income categories and geography as measured in tax returns. The direction of causality between income and health is murky, which is one reason that researchers often define SES using education rather than income. But we believe the connection between age-related transfers of in-kind and cash assistance on the one hand and older-age health outcomes on the other are likely to be less cloudy, based on earlier findings drawn from experiences in other industrialized countries. In this study we assess these across the SES gradient in the U.S. Of particular interest is the expansion of U.S. old-age health benefits in the form of Medicare Part D, the prescription drug benefit that was introduced in 2006. Although we expect to find increases in age-related transfers by SES to be associated with improvements in health, we also believe that the exposition of trends in transfers and health by SES that we may find will prove revealing.

Preliminary studies thus far have focused on measuring SES among respondents aged 65 and over in the 2000 and 2010 waves of the HRS, and on examining how average health measures vary across SES groupings at those ages. The investigation team found that quintiles of educational attainment among this age group may have shifted upward by as much as one year of schooling over this decade, which is also roughly the increase in average education and likely reflects the expansion in attainment for cohorts born around midcentury. The key methodological challenge we face is in assessing whether education, our preferred index of SES, should be measured in absolute levels, as is common in some literatures, or in percentiles. At least for average health outcomes during this period, early results suggest that defining SES in absolute instead of relative terms appears to matter somewhat for the magnitudes and sometimes even for the signs of measured trends. The team also found that trends in the SES gradient in health depended on the particular measure of health, with results most notably different between self-reported health and disability indexes. Results also depended on whether one examined the entire gradient or instead focused on either the high end or the low end. Because health is multidimensional, these findings are perhaps to be expected, if methodologically challenging. Further work awaits the reconstruction of the U.S. NTA by SES.

Project Director: 
Ronald D. Lee (Demography, UC Berkeley) and Ryan D. Edwards