Economy

Poverty in America

1959— present

The War on Poverty cut the rate sharply in a decade — then it barely moved for fifty years.

The U.S. has measured its official poverty rate since 1959, when more than one in five Americans were poor. The 1960s War on Poverty and the growth of the safety net drove a sharp early drop, but the official rate has hovered in a narrow band ever since — even as the measure itself became contested. This traces the long record, how poverty is measured (and why the newer Supplemental measure tells a different story), who's poor and why, what the safety net does, and why a rich country still has the poverty it does. A different question than Inequality: absolute hardship vs the shape of the distribution.

Official rate, 1959 → 2024

22.4→10.6

% below the threshold · 35.9M people in 2024 (Census hstpov2, earliest year 1959)

Child poverty, 2021 → 2022

5.2→12.4

% on the SPM · more than doubled the year the expanded Child Tax Credit lapsed

Lifted out by Social Security

28.7M

in 2024 — the largest antipoverty program, by a factor of four (Census P60-287)

Below half the threshold

5.0%

2024 deep poverty, up from 3.7% in 1975 — the earliest year Census publishes

Black vs White non-Hispanic

18.4/7.6

% in 2024 · a 10.8-point gap, down from 37.0 points in 1959

The short answer

The American poverty line is the cost of the U.S. Department of Agriculture's cheapest emergency food plan in 1963, multiplied by three, and adjusted for inflation every year since. That is the whole formula. Mollie Orshansky, the Social Security Administration analyst who built it, used a multiplier of three because a 1955 survey found families spent about a third of their after-tax income on food. Families now spend closer to an eighth. The multiplier has never changed.

Two official numbers for the same year · 1959 — 2024

Official measure (from 1959)Supplemental Poverty Measure (from 2009)

Census hstpov2 for the official measure, whose first year is 1959, and P60-287 Table B-2 for the SPM, whose first year is 2009 — that is the entire published series, not a chosen window. The two lines are not a before/after: they are two measures of the same years, and where they overlap the SPM is usually the higher of the two. Three SPM years (2013, 2017, 2019) have two published values because of methodology changes; the value shown is the one comparable with the year that follows.

The two measures, side by side

Official measure

Supplemental (SPM)

Time series

1959 — present

2009 — present

2024 rate

10.6%

12.9%

Threshold from

Three times the cost of a minimum food diet in 1963

Recent spending on food, clothing, shelter, utilities, phone, and internet

Adjusts for

Family size, composition, age of householder

Size, composition, housing tenure, and local housing costs

Updated by

CPI-U, all items

Five-year moving average of spending on that basket, lagged a year

Counts as income

Gross pretax cash only

Cash plus SNAP, housing, school lunch, WIC, energy aid; minus taxes, work expenses, medical costs, child support paid

Sharing unit

People related by birth, marriage, or adoption

That, plus unmarried partners, their relatives, coresident and foster children

Used for

Official counts; program eligibility formulas; funding allocations

Research. It has no statutory or administrative role.

Where the line actually sits · 2024

One person under 65$16,320

One person 65 or over$15,050

Two people$20,220

Three people$24,950

Census weighted-average thresholds, hstpov1. These are national figures: the official measure applies the same line in Manhattan and rural Mississippi. The SPM does not, which is most of why the two measures disagree about which states are poorest.

Why the argument never ends

Both National Academies panels asked to review the measure — in 1995 and again in 2023 — concluded it was not fit for purpose. It survives anyway, for a reason that has nothing to do with statistics: hundreds of programs and funding formulas are keyed to it, so any revision reassigns real money between states, programs, and people.

The SPM was the compromise. It fixes nearly everything the 1995 panel objected to — it counts SNAP and tax credits as income, subtracts medical and work costs, and adjusts for local housing — and in exchange it was given no official status at all. Census publishes both numbers every September and is careful to say the SPM "does not replace the official poverty measure."

A third answer, from outside government

Columbia's Center on Poverty and Social Policy reconstructed the SPM back to 1967, which government data cannot do. On their historical SPM, poverty fell from 18.5% in 1967 to 12.9% in 2024. On an SPM anchored to 2012 living standards — an absolute line, moved only by inflation — it fell from 25.8% to 10.1%, a 61% decline.

The same analysis produces the sharpest finding of all. Strip out taxes and transfers, and the 2024 poverty rate would be higher than in 1967: 23.7% against 21.9%. Market outcomes did not improve. Government transfers, which cut the rate by 16% in 1967 and 46% in 2024, are the entire story.