Take the 3,142 US counties with figures for both the share of residents in work and the share living in poverty. Across those counties, the two usually run in opposite directions: where more people are in the labour force, poverty tends to be lower. Usually, but loosely. 802 counties, about 26% of the country and 33% of its population, sit in the two corners where the pattern inverts. In 401 of them, a high share of people work, and poverty is still high. In another 401, few people work, and poverty is low anyway.
The median county has 58.6% of residents in the civilian labour force and 13.2% living below the poverty line. The 401 counties above the median on both hold 27% of the US population, so this is not a rural curiosity. Los Angeles County reports 64.8% in work and 13.6% in poverty. Kings County, which is Brooklyn, reports 64.1% and 18.9%. Davidson County, which is Nashville, has one of the higher work rates in the country at 71.9%, and poverty of 13.9%, still above the county median.
At the far end of the same corner, the gap is wider. Ziebach County in South Dakota reports 61.0% of residents in the labour force, above the median county, and 38.2% in poverty, with a median household income of $47,333. Tunica County in Mississippi reports 60.9% in work and 33.8% in poverty. In these places, the number of people working is not the constraint. What the work pays is.
The opposite corner holds 401 counties and 6% of the population. Their median household income is $66,341, close to the national county middle, and the thing that separates them is age: the median county in this group has 23.6% of residents aged 65 and over, against 17.0% in the busy, higher-poverty corner. Sumter County in Florida is the clearest case at any size, with 25% of residents in the labour force, the lowest of any county with more than 25,000 residents and a long way below the next one at 39.5%. It has 57.8% of residents aged 65 and over, the highest share of any county of any real size, and poverty of 9.1%. Pensions, savings, and social security are income that does not come from a job.
The share of people in work is often used as a stand-in for how well a place is doing, and for most counties it earns that. Pulling work and poverty together for every county in the country, then keeping both current, is the part that used to be slow and expensive, so treating one as a proxy for the other was a reasonable way to work.
With both figures side by side, the two corners change what a plan should say. Screen counties on work alone and the 401 counties where people work and are still poor read as healthy, which for a service, a wage assumption, or a campaign about affordability is exactly backwards. Screen on the same measure and the 401 older counties read as struggling, when their households are on incomes that arrive whatever the labour market does. Those two groups hold about a third of the country between them, and they need opposite conclusions drawn about them.
Cambium AI builds synthetic populations from this same public data, so a research, planning, or marketing team can hold work, income, poverty, and age together for the counties they cover and see which corner each one is in before the plan is written. These figures come from the American Community Survey, the Census Bureau's annual household survey, which reports all four for every county in the country.
Data source: U.S. Census Bureau, American Community Survey (ACS) 5-Year Estimates