Mortality Change among Less Educated Americans

Measurements of mortality change among less educated Americans can be biased because the least educated groups (e.g., dropouts) become smaller and more negatively selected over time. We show that mortality changes at constant education percentiles can be bounded with minimal assumptions. Middle-age mortality increases among non-Hispanic Whites from 1992 to 2018 are driven almost entirely by the bottom 10 percent of the education distribution. Drivers of mortality change differ substantially across groups. Deaths of despair explain most of the mortality change among young non-Hispanic Whites, but less among older Whites and non-Hispanic Blacks. Our bounds are applicable in many other contexts. (JEL I12, I26, J15)