Metropolitan Age Structure in the United States

We live for just these twenty years
Do we have to die for the fifty more?

— David Bowie, “Young Americans,” 1975

This is an attempt to build a typology of age structures throughout the United States, covering 387 metropolitan statistical areas and 3,144 counties. While median age is the usual population statistic used to assess whether an area is comparatively young or old, this is a poor way of distinguishing between places with radically different characteristics in terms of age distribution and aging-in-place. This analysis provides the tools for understanding what makes a high-fertility exurb, a garrison, a college town, and a working-age magnet economy different from one another.

By Kevin Contino · July 2026 · Every figure is rebuilt from a cached source file by a make all pipeline. The full technical report and the interactive metro map carry the detail.

› Open the interactive map

Data source

39.38 national median age, years, 7/1/2025
190 / 387 metros below it
2020–2025 estimates covering 4/1/2020 to 7/1/2025

Census Bureau Vintage 2025 population estimates, released June 2026. Every cached file is recorded with its source URL, byte size, SHA256 and retrieval timestamp in the source manifest.

Age structure has shapes, not just averages

Six single-year age pyramids, one per typology cluster: Oklahoma City, Watertown-Fort Drum, Naples-Marco Island, Ann Arbor, Morristown and Jefferson City.
Figure 1. The confounds are visible as shapes. Single-year age pyramids, one exemplar per cluster: Watertown–Fort Drum's spike at 18–24 is the eponymous Fort Drum garrison; Ann Arbor's is a Big Ten college; Naples–Marco Island is a real-life version of the “Quietus” from Children of Men. All figures are drawn on a shared x-scale as a percentage of each metro's population. Ages 0–84 are true single years; the top-coded 85-and-over bin is omitted from the bars. Group quarters are retained here, because institutional bulges are among the population features that distinguish the clusters from one another.

Nine shape-based features per metro help to define their population types: modality count, a dissimilarity index, group-quarters share by type, sex ratio at 18–29, and the under-18 and 65+ shares. College towns, retirement destinations and aged-in-place metros each come out cleanly.

However, high-fertility metros, border and agricultural metros, and prime-age magnets collapse into a single shape-based cluster, and require a separate instrument to recognize.

Cohort progression

That separate instrument is the cohort change ratio: if we follow one birth cohort between two dates, we can see whether the county retains a cohort over time or churns through. This analysis looks at the cohort change ratio by comparing each area's 25–29 population as of 2020 to its 30–34 population as of 2025: CCR = P(30–34, 7/1/2025) ÷ P(25–29, 7/1/2020).

Above 1.0 the county gained that cohort; below 1.0 it lost it. It is age-specific and requires no assumption about where migrants came from.

Two US maps of the cohort change ratio for the 25-29 cohort of 2020: a county choropleth above, and below the same data as a dot cartogram with dot area proportional to county population.
Figure 2. The Sun Belt and the mountain west gained; much of the interior did not. Diverging orange–blue scale centred at 1.0, both arms generated at matched lightness so equal gains and losses read equally strongly. The span is the 90th percentile of |CCR − 1|, which clips the most extreme 10% of counties into the end steps. The upper panel greys counties under 20,000 population, whose ACS-derived inputs carry larger margins of error. The lower panel is the population-weighted view.

Composite magnet index

A metro can be young for reasons that have nothing to do with anybody choosing it. Births make a metro young. So does a barracks, or a campus, or a farm labour force. Only one of the four is a decision an adult made about where to live, and it is the only one most people mean when they call a place a magnet.

The magnet index measures that directly, and it asks three questions of every metro. Does it keep the people who arrive in their late twenties and thirties, or do they leave again by their mid-thirties? Are jobs in that age band growing there? And when Americans move between metros, does the flow run toward this one or away from it? A metro scoring high is winning working-age adults from the rest of the country. A metro scoring low is losing them, whatever its median age says.

A US metro map of the composite magnet index alongside a dot plot of the top 30 metros ranked, with excluded metros marked as diamonds.
Figure 3. The composite index, and the metros an exclusion criterion removes. A weighted mean of percentile ranks: mean cohort change ratio across the 25–39 cohorts (0.35), early-career share (0.15), net domestic migration rate (0.15), QWI employment growth in the 25–34 band (0.20), and inverted old-age dependency (0.15). Diamonds mark metros excluded by at least one criterion — group quarters, a top-decile under-5 share, or international migration exceeding half of net migration.

Why some metros are young

Stacked bars decomposing the cause of youth into fertility, institution, immigration and attraction for the 50 metros with the lowest median age.
Figure 4. Four causes of a low median age, separated. Each of the 190 metros below the national median age is assigned a dominant cause; the 50 youngest are shown, youngest at the top. Cause scores are means of percentile ranks computed within the below-median set, then normalised to sum to one.

Methods

Everything above is produced by make all: fetch (cached, with a manifest), build, analyse, validate, figures, interactive map, report. Mapping is Albers Equal Area (EPSG:5070) with Alaska (EPSG:3338) and Hawaii (EPSG:26962) reprojected, rescaled and inset. Additional details are in the technical report.