What the study measured
The 2016 JAMA paper by Raj Chetty and colleagues examined the association between income and life expectancy in the United States from 2001 through 2014. The researchers combined 1.4 billion deidentified tax records with Social Security Administration death records. They estimated race- and ethnicity-adjusted life expectancy at age 40 by household income percentile, sex, time, and geography.
The exposure was pretax household earnings. The result is a set of population estimates, not a rule that maps an individual bank balance to a death date. The project publishes its formatted tables, codebooks, and replication materials through the Health Inequality Project.
The model’s one selection
Life Days Left uses the national, race-adjusted expected age at death at age 40 for the 75th household-income percentile. It averages the published male value, 83.785507 years, and female value, 86.37175 years, producing an unisex baseline of 85.0786285 years.
The 75th percentile is a fixed model setting. It was chosen to make the public chart deterministic and to state the income assumption instead of hiding it inside a generic national average. It is not a claim that the chart audience is at that percentile, and the site never asks for or estimates a visitor’s income.
Association is not causation
The study documents a relationship between income and longevity, but its design does not show that transferring income to a person will add a specified number of years to that person’s life. Education, health behavior, neighborhood, occupation, access, and other measured or unmeasured factors can be related to both income and mortality.
That caution belongs in the model interpretation. “Income-linked baseline” is accurate; “income causes this lifespan” is not. The chart also omits the geographic variation that was central to part of the original research. It should not be used to compare cities, evaluate policy, or recommend financial or medical decisions.
How the baseline is used—and not used
After selecting the baseline, the model applies only a relative birth-year adjustment from the SSA 2026 cohort series. It does not refresh the Chetty estimate with current earnings data, and it does not preserve separate male and female outputs. Those simplifications make one compact generation chart possible, while discarding information that would matter in individual or demographic analysis.
Readers who need the exact transformation can inspect the methodology. Readers who need the current chart numbers can use the data page, which states the as-of date and model version beside the values.