Rich Americans live about as long no matter where they live. Poor Americans' lifespans depend on their address — and the places that stretch them are not the places with the most doctors or insurance, but the places where fewer people smoke.
In 2016, Raj Chetty's team linked 1.4 billion tax returns to Social Security death records and computed something no survey could: the life expectancy of Americans at every income level, in every county. Two findings organize everything below. First, the rich live longer — about six and a half years longer, comparing the top quarter of household incomes to the bottom. Second, and stranger: the first fact is geographically stable and the second is not. Being rich buys roughly the same lifespan in Detroit as in Manhattan. Being poor does not.
That asymmetry is this book's thesis in a single dataset. If place merely reflected the people in it, the poor would fare equally badly everywhere. Instead, place does its work most powerfully on those with the least — the same pattern the Moving to Opportunity experiment found for children.
Life expectancy here means expected age at death for a 40-year-old, race-adjusted, averaged over 2001–2014. Each step up the income ladder buys years. It also closes the gender gap: among the poorest quarter, women outlive men by 5.3 years; among the richest, by 2.3. Most of the income-longevity gradient in America is a gradient among men.
Each dot is one large county (250,000+ residents). Across the horizontal axis — the lifespan of its richest quarter — the dots barely spread: 2.5 years from the 5th to the 95th percentile county. Across the vertical axis — the lifespan of its poorest quarter — they spread 4.1 years, half again as much, from Texas border counties and the post-industrial Midwest at the bottom to New York City's boroughs at the top. For the rich, geography is a rounding error. For the poor, it is several years of life.
Color every measurable county by how long its poorest quarter lives and the geography is unmistakable: long lives for the poor along the coasts and in immigrant gateways; short ones in the eastern industrial belt, Appalachia's edge, the Gulf Coast of Texas, and parts of the interior West.
The ten best big counties for a poor person's lifespan include four New York City boroughs and a string of high-immigration coastal counties. The worst are Gulf-coast Texas, Flint's and Detroit's counties, and the urban industrial Midwest — places where a poor 40-year-old gives up five to six years relative to Queens.
Correlate the poor's local lifespan with everything measurable about a county and a sharp hierarchy appears. Health behaviors dominate: where fewer poor residents smoke, the poor live longer (r = −0.57). Local affluence and education matter. But the variables the policy debate fixates on barely register: the uninsured share, Medicare spending and quality, unemployment, even local poverty itself all sit near zero. The rich quartile's bars (hollow) are flatter for behaviors — their lifespan travels with them; the poor's is negotiated locally.
These are correlations across 1,559 counties, not causal estimates; poor households who value health may also move to places like Santa Cruz. The immigrant share correlates strongly (r = 0.66) partly because immigrants themselves live exceptionally long, and they are overrepresented among the urban poor. And the period is 2001–2014 — before the opioid crisis fully reshaped the geography of death among the poor.
But the central asymmetry survives every caveat, and it carries the chapter's freight: in America the question "how long do the poor live?" has no national answer. It is answered county by county — by smoking norms, food, housing, and the accumulated local texture of life among people with no buffer against any of it. The rich are insured against their address. The poor are exposed to it.